Download Running head: The evolutionary genetics of personality

Running head: THE EVOLUTIONARY GENETICS OF PERSONALITY
The evolutionary genetics of personality
Lars Penke
Humboldt University, Berlin
and
International Max Planck Research School LIFE, Berlin
Jaap J. A. Denissen
Utrecht University
Geoffrey F. Miller
University of New Mexico
Total word count:
19,975
Keywords:
Evolutionary genetics, personality psychology, behaviour
genetics, molecular genetics, evolutionary psychology,
personality traits, intelligence, psychopathology, geneenvironment interactions, personality structure
Corresponding author:
Lars Penke, Humboldt University, Institute of Psychology, Rudower Chaussee 18, D-12489
Berlin. Email: [email protected]
The evolutionary genetics of personality
2
Abstract
Genetic influences on personality traits are ubiquitous, but their nature is not well
understood. A theoretical framework might help, and can be provided by evolutionary
genetics. We assess three evolutionary genetic mechanisms that could explain genetic
variance in personality differences: selective neutrality, mutation-selection balance, and
balancing selection. Based on evolutionary genetic theory and empirical results from
behaviour genetics and personality psychology, we conclude that selective neutrality is
largely irrelevant, that mutation-selection balance seems best at explaining genetic variance
in intelligence, and that balancing selection by environmental heterogeneity seems best at
explaining genetic variance in personality traits. We propose a general model of heritable
personality differences that conceptualises intelligence as fitness components and
personality traits as individual reaction norms of genotypes across environments, with
different fitness consequences in different environmental niches. We also discuss the place
of mental health in the model. This evolutionary genetic framework highlights the role of
gene-environment interactions in the study of personality, yields new insight into the personsituation-debate and the structure of personality, and has practical implications for both
quantitative and molecular genetic studies of personality.
(180 words)
The evolutionary genetics of personality
3
Evolutionary thinking has a long history in psychology (e.g. James, 1890; McDougall, 1908,
Thorndike, 1909). However, the new wave of evolutionary psychology (e.g. Buss, 1995;
Tooby & Cosmides, 2005) has focused almost exclusively on human universals – the
complex psychological adaptations that became genetically fixed throughout our species due
to natural selection (Andrews, Gangestad & Matthews, 2002) and that should therefore show
zero genetic variation and zero heritability (Tooby & Cosmides, 1990). In sharp contrast, one
of personality psychology’s most important findings in the last three decades has been that
virtually every aspect of personality is heritable (Plomin, DeFries, McClearn & Mc Guffin,
2001). This fact is now so well established that Turkheimer (2000; Turkheimer & Gottesman,
1991) even called it a law. The mismatch between evolutionary psychology’s adaptationist
focus on human universals and the omnipresence of heritable variance in human personality
might explain why early approaches towards an evolutionary personality psychology (Buss,
1991; Tooby & Cosmides, 1990; MacDonald, 1995, 1998) remained rather unsatisfactory
(see Miller, 2000a; Nettle, 2006a).However, traditional behaviour genetics did not explain the
evolutionary origins and persistence of genetic variation in personality, and sometimes even
viewed genetic variation in traits as evidence of their evolutionary irrelevance. Thus, the
evolutionary psychology of human universals and the behaviour genetics of personality
differences share a biological metatheory, but had almost no influence on each other (Tooby
& Cosmides, 1990, 2005; Plomin et al., 2001).
We believe that this mutual neglect has been unfortunate for both fields, and has
especially harmed the development of an integrative evolutionary personality psychology.
Evolutionary studies of species-typical universals and individual differences were already
successfully merged during the ‘Modern Synthesis’ in the 1930s, when Sir Ronald A. Fisher,
Sewell Wright, J. B. S. Haldane, and others united the branches of biology that were founded
by the cousins Charles Darwin (the father of adaptationism) and Sir Francis Galton (the
father of psychometrics and behaviour genetics) (Mayr, 1993). These 1930s biologists
created what is now known as ‘evolutionary genetics’, which deals with the origins,
maintenance, and implications of natural genetic variation in traits across individuals and
The evolutionary genetics of personality
4
species. Evolutionary genetics mathematically models the effects of mutation, selection,
migration, and drift on the genetic basis of traits in populations (Roff, 1997, Maynard Smith,
1998). In the following, we will argue that personality psychology needs an evolutionary
genetic perspective in order to draw maximal benefits from behaviour genetic findings and
the evolutionary metatheory. This is important, since understanding the evolutionary
behaviour genetics of personality is fundamental to the future development of a more unified
personality psychology (McAdams & Pals, 2006).
Overview
The central topic of this review is how evolutionary genetics can inform our theoretical
understanding of heritable personality differences and their genetic foundations. We use
‘personality differences’ in the broad European sense of encompassing individual differences
in both cognitive abilities and personality traits (e.g. Eysenck & Eysenck, 1985). Cognitive
abilities reflect an individual’s maximal performance in solving cognitive tasks. It is wellestablished that a single continuum of general intelligence (g), ranging from mild mental
retardation to giftedness, explains a large proportion of the individual differences in cognitive
abilities across domains (Jensen, 1998), especially on genetic level (Plomin & Spinath,
2004). Our discussion on cognitive abilities will be focused on this general intelligence
dimension. Personality traits reflect an individual’s set of typical behavioural tendencies
exhibited in situations that leave room for diverse adaptive responses. The myriad of
personality trait dimensions are usually organized in structural models. Broad personality trait
domains, as in the Five Factor Model of personality (FFM), are generally regarded as stable
and temperamental in nature (John & Srivastava, 1999). They are what we mean by
‘personality traits’.
We argue that the classical distinction between cognitive abilities and personality
traits is much more than just a historical convention or a methodological matter of different
measurement approaches (see Cronbach, 1949), and instead reflects different kinds of
selection pressures that have shaped distinctive genetic architectures for these two classes
The evolutionary genetics of personality
5
of personality differences. In order to make this argument, we will first give a brief
introduction to the nature of genetic variation and the major mechanisms that contemporary
evolutionary genetics proposes for its maintenance in populations. After this, we will critically
review earlier evolutionary approaches to personality and clarify the role of environmental
influences within this approach. This will culminate in an integrative model of the evolutionary
genetics of personality differences, including new, theory-based definitions of cognitive
abilities and personality traits, as well as a discussion of how common psychopathologies
(such as schizophrenia and psychopathy) may fit into an evolutionary genetic model of
personality differences. Finally, we will discuss this model’s implications for an integrated
evolutionary personality psychology grounded in both behaviour genetics and evolutionary
genetics.
What is Genetic Variation?
Most personality psychologists now accept Turkheimer’s (2000) first law of behaviour
genetics (‘everything is heritable’). Yet how does systematic genetic variation in personality
traits arise? A complete understanding of the insights offered by evolutionary genetics
requires a brief review of some of the basics of genetics and evolutionary theory, which we
provide in the following.
The Human Genome. The human genome consists of about 3.2 billion base pairs that
are unequally spread across 23 chromosomes. Only about 75 million (2.3%) of these base
pairs are organized in roughly 25,000 genes (i.e. regions or ‘loci’ translated into actual
protein structures); the rest (traditionally called ‘junk DNA’) do not code for proteins, but may
play important roles in gene regulation and expression (Shapiro & von Sternberg, 2005). On
average, any two same-sex individuals randomly drawn from the total human population are
99.9% identical with regard to their base pairs (Human Genome Project, 2001), even though
genomic identity is somewhat further attenuated by copy-number variations (CNVs, individual
differences in the repetitions of DNA segments) (Redon et al., 2006). This species-typical
genome contains the universal human heritage that ensures the highly reliable ontogenetic
reoccurrence of the complex functional human design across generations (‘design
The evolutionary genetics of personality
6
reincarnation’, Barrett, 2006; Tooby, Cosmides & Barrett, 2005). Adaptationistic evolutionary
approaches usually care only about this universal part of the genome and its species-typical
phenotypic products (Andrews et al., 2002; Tooby & Cosmides, 2005).
Mutation. During an individual lifespan, the genome is passed from mother cells to
daughter cells by self-replication, and if this results in a germline (sperm or egg) cell, half of
the genome eventually ends up combining with an opposite-sex germline cell during sexual
reproduction, and is thus passed from parent to offspring. While genomic self-replication is
astonishingly precise, it is not perfect. Replication errors can occur in the form of point
mutations (substituting one of the four possible nucleotides in a base pair for another one,
also referred to as single nucleotide polymorphisms (SNPs)), CNVs (duplications or deletions
of base pair sequences), or rearrangements of larger chromosomal regions (e.g.
translocations, inversions). All of these copying errors are referred to as mutations, and they
are ultimately the only possible source of genetic variation between individuals. Recent
scans of whole human genotypes reported 9.2 million candidate SNPs (International
HapMap Consortium, 2005) and 1,447 candidate CNV regions (Redon et al., 2006).
Sexual reproduction endows an individual with a unique mixture of their parents’
genotypes. In the short term, this process of sexual recombination is the major cause of
genetic individuality. In the evolutionary long term, however, sexual recombination is less
important, since it just reshuffles the parental genetic variation that was once caused by
mutation. By convention, mutations that continue to be passed on to subsequent generations
and that reach an arbitrary threshold of more than 1% prevalence in a population are called
‘alleles’. Since all alleles are mutations, we regard this distinction as hardly helpful. In
contrast, ‘polymorphism’ is a more neutral term for genetic variants that can be at any
prevalence. In order to highlight the evolutionary genetic perspective, we will use the terms
‘mutation’ and ‘polymorphism’ interchangeably.
Some mutations are phenotypically neutral, often because they do not affect protein
structure or gene regulation. Most mutations in protein-coding and genomic regulatory
regions, however, tend to be harmful to the organism because they randomly disrupt the
The evolutionary genetics of personality
7
evolved genetic information, thereby eroding the complex phenotypic functional design
(Ridley, 2000; Tooby & Cosmides, 1990). Only very rarely does a random mutation improve
the functional efficiency of an existing adaptation in relation to its environment, which is more
likely if the environment has changed since the adaptation evolved (Brcic-Kostic, 2005).
Deletions, insertions, and larger rearrangements of base pair sequences tend to have quite
strong disruptive effects on the phenotype, often leading to prenatal death or severe birth
defects. Point mutations (SNPs) and duplication-type CNVs (see Hurles, 2004), on the other
hand, can have phenotypic effects of any strength, including quite mild effects, and it is likely
that they are the most common source of genetic variation between individuals.
Behaviour Genetics. Quantitative traits, such as intelligence and personality traits, are
polygenic - they are affected by many mutations at many genetic loci, each of which is called
a quantitative trait locus (QTL) (Plomin, Owen & McGuffin, 1994). Quantitative behaviour
genetics basically compares trait similarities across individuals that systemically differ in the
genetic or environmental influences they have in common (e.g. identical vs. fraternal twins,
adoptive vs. biological children), to decompose the variation of quantitative traits, and their
covariances with other traits, into genetic and environmental (co)variance components. It
also tries to estimate how much of the genetic (co)variance is due to ‘additive effects’ of
QTLs (which allow traits to ‘breed true’ from parents to offspring) versus interactions between
alleles at the same genetic locus (‘dominance effects’) or across different genetic loci
(‘epistatic effects’). Dominance and epistatic effects lead to non-additive genetic variance
(VNA) between individuals, as opposed to the additive genetic variance (VA) caused by
additive effects. Together with the environmental variance (VE) and gene-environment (GxE)
interactions, these components determine the phenotypic variance (VP) that we can observe
in personality differences. In contrast to quantitative behaviour genetics, molecular behaviour
genetics uses so-called ‘linkage’ and ‘association’ methods to directly analyse human DNA
variation in relation to personality variation, to identify the specific QTLs that influence
particular trait (co)variations (Plomin et al., 2001).
The evolutionary genetics of personality
8
Natural selection. Mutations in functional regions of the genome provide half of the
basic ingredients for biological evolution. The other half is natural selection, which is the
differential reproduction of the resulting phenotypes (Darwin, 1859). Any mutation that affects
the phenotype is potentially visible to natural selection, though to varying degrees. Of course,
those rare mutations that actually increase fitness will tend to spread through the population,
driving adaptive evolution. Selection is most obvious against mutations that lead to
premature death or sterility. Such mutations are eliminated from the population within one
generation, and can only be reintroduced by new mutations at the same genetic loci.
Mutations with less severe effects tend to persist in the population for some time; they are
selected out of the population more quickly when their additive effect reduces the fitness of
the genotype (i.e., its statistical propensity for successful reproduction) more severely. This
relationship between the additive phenotypic effect of a genetic variant and its likely
persistence in a population is described by the fundamental theorem of natural selection
(Fisher, 1930).
To summarize, any genetic variation in any human trait is ultimately the result of
mutational change in functional regions of the species-typical genome. Natural selection
counteracts disruptive changes by eliminating harmful mutations from the population, at a
rate proportional to the mutation’s additive genetic reduction in fitness. Only mutations that
affect the organism’s fitness in a positive or neutral way can spread in the population and will
reach the 1% prevalence of an ‘allele’. Most psychological traits, including personality
differences, are complex in design and continuously variable across individuals, indicating
that many polymorphisms at many loci are responsible for their genetic variation.
Why is There Genetic Variation in Personality?
Also else being equal, it seems plausible that natural selection should favour an
invariant, species-typical genome that codes for a single optimal phenotype with optimal
fitness. In other words, evolution should eliminate genetic variation in all traits, including all
aspects of personality. So how can personality differences still be heritable (i.e., genetically
The evolutionary genetics of personality
9
variable) after all these generations of evolution? To answer this fundamental question, an
evolutionary genetic approach to personality is needed.
With the growing acceptance of evolution as a metatheory for psychology, more and
more personality psychologists are trying to conceptualize personality in an evolutionary
framework. Unfortunately, these good intentions seldom lead to more than an affirmation that
certain heritable dimensions are part of our evolved human nature (e.g. McCrae & Costa,
1996; Ashton & Lee, 2001; McAdams & Pals, 2006). Even worse, some conceptualisations
of human cognitive abilities ignore genetic variation completely and discuss these heritable,
variable traits as if they were invariant adaptations (e.g. Cosmides & Tooby, 2002;
Kanazawa, 2004). Other authors (Goldberg, 1981; Buss, 1990; Hogan, 1996; Ellis, Simpson
& Campbell, 2002) take genetic variation in personality differences for granted, and try to
understand evolved features of our ‘person perception system’ that explain why we
categorize others along these dimensions. Few have attempted an evolutionary genetic
approach to explain the persistence of heritable variation in personality itself.
Evolutionary genetics offers a variety of mechanisms that could explain persistent
genetic variation in personality differences. These mechanisms include selective neutrality
(where mutations are invisible to selection), mutation-selection balance (where selection
counteracts mutations, but is unable to eliminate all of them), and balancing selection (where
selection itself maintains genetic variation). Recent theoretical developments make it
possible to predict how each of these mechanisms would influence certain genetic and
phenotypic features of traits (see Table 1). Conversely, if these features are known for a
given trait, it is possible to identify which evolutionary processes likely maintained the genetic
variants that underlie its heritability. We will now review existing attempts to explain
personality differences from an evolutionary perspective, and evaluate them in the light of
modern evolutionary genetics.
Insert Table 1 about here
Can Selective Neutrality Explain Genetic Variance in Personality?
The evolutionary genetics of personality
10
Tooby and Cosmides (1990) developed an early and highly influential perspective on
the evolutionary genetics of personality. They reviewed the state of evolutionary genetics at
that time, but, as major advocates of an adaptationistic evolutionary psychology, they
focused on species-typical psychological adaptations and downplayed genetic variation as
minor evolutionary noise. In their view, one plausible mechanism that could maintain genetic
variation in psychological differences is selective neutrality (Kimura, 1983). This occurs when
fitness-neutral mutations (that have no net effect on survival or reproductive success,
averaged across all relevant environments) accumulate to increase genetic variance in a
trait. For example, the exact route that the small intestine takes within one’s abdomen may
have little influence on digestive efficiency, so neutral genetic variation that influences
patterns of gut-packing could easily accumulate. In the evolutionary short-term, selective
neutrality allows genetic variance in traits to increase.
However, what happens in the evolutionary long-term to selectively neutral traits?
Since neutral mutations are, by definition, unaffected by natural selection, the only
evolutionary force that can affect neutral genetic variation is genetic drift – and drift always
tends to decrease genetic variance. Drift is basically the fixation (to 100% prevalence) or
elimination (to 0% prevalence) of a polymorphism by chance. There is only one factor that is
known to be important for the efficacy of drift: it is stronger when the ‘effective population
size’ (Ne) (the average number of reproductively active individuals in a population) is smaller
(Lynch & Hill, 1986). What is really critical for the effect of genetic drift is the minimum N e
during occasional harsh conditions (e.g. ice ages, disease pandemics) that created ‘genetic
bottlenecks’ (especially small effective population sizes). In humans, 10,000 seems to be a
good estimate for the minimum Ne (Cargill et al., 1999). Mathematical models show that, with
such a relatively large Ne, drift is fairly weak and selective neutrality could, in principle,
account for almost all genetic variance in any human trait (Lynch & Hill, 1986).
So far, so good: perhaps most genetic variation in human personality is due to
selective neutrality – maybe there is no average net fitness cost or benefit to being
extraverted versus introverted, or agreeable versus egoistic. However, the critical
The evolutionary genetics of personality
11
assumption for selective neutrality is that genetic drift is more important than natural
selection in affecting a trait’s genetic variance. This is only the case if the selection coefficient
s is less than about 1/4Ne (Keller & Miller, 2006a). Thus, the larger the effective population
size, the harder it is for a trait to be selectively neutral. Given the reasonably large estimate
of minimum human Ne from above (10,000), a typical human trait is selectively neutral only if
the average net fitness of individuals with a certain polymorphism is between 99.997 and
100.003% of the average fitness of individuals without that polymorphism (Keller & Miller,
2006a). For example, an allele that influences extraversion would be truly neutral only if
extraverts had, not just the same number of 1st-generation offspring as introverts, but
(almost) exactly the same average number of 15th generation descendants (great13
grandchildren). In addition, this finely-balanced neutrality must hold across all relevant
environments: if there are some environments in which outgoing, risk-seeking extraverts do
better, and other environments in which shy, risk-averse introverts do better (a GxE
interaction), then extraversion would be under balancing selection (see below), not selective
neutrality.
This makes selective neutrality an implausible explanation for heritable personality
differences, because human personality traits influence outcomes in all areas of life (Ozer &
Benet-Martinez, 2006), including such obviously fitness-relevant aspects as health
(Neeleman, Sytema & Wadsworth, 2002), life expectancy (Friedman et al., 1995), mating
strategies (Nettle, 2005), and reproductive success (Eaves et al., 1990). Indeed, similar nonneutral relationships between personality and fitness have been observed in various other
species (Dingemanse & Réale, 2005). The relation between cognitive abilities and fitness
components has also been impressively demonstrated by Gottfredson (2004, in press),
Deary (e.g. Deary et al., 2004; Deary & Der, 2005), and Miller (2000b; Prokosch, Yeo &
Miller, 2005).
How could we tell if a heritable individual difference was the outcome of selective
neutrality? Typically, selective neutrality leads to a distinct structure of genetic variation in
quantitative traits (such as personality differences). If a mutation affects the phenotypic
The evolutionary genetics of personality
12
expression of a trait, it will first of all have a main effect, which means it will contribute to the
additive genetic variance (VA) of the trait. Only if the mutation happens to interact with other
polymorphisms (at the same or other loci, through dominance or epistasis, respectively), will
it contributes to the non-additive genetic variance (VNA) of the trait. This is exactly the same
logic that holds for any statistical analysis: ceteris paribus, main effects are much more likely
than interaction effects. Since all else is equal under selective neutrality by definition, we can
expect low absolute values of VNA for any selectively neutral trait (Lynch & Hill, 1986; Merilä
& Sheldon, 1999), and a very small proportion of non-additive genetic variance (Dα), defined
by Crnokrak and Roff (1995) as:
Dα = VNA / (VNA + VA).
(1)
Traits with a recent history of selection, by contrast, should show a significant absolute and
proportional amount of VNA (Crnokrak & Roff, 1995; Merilä & Sheldon, 1999; Stirling, Réale &
Roff, 2002). This follows from Fisher’s (1930) fundamental theorem of natural selection:
since VA is passed directly from parents to offspring, it will be reduced very quickly by natural
selection for any non-neutral trait. VNA, on the other hand, is affected much more weakly by
selection, since the interacting genetic components that constitute the VNA are continuously
broken apart by sexual recombination and thus not passed from parents to offspring. As a
result, a high proportion of VNA in a trait would argue against the trait’s selective neutrality.
There is now strong evidence that personality traits show substantial VNA (Eaves et al., 1998;
Keller et al., 2005) - including some initial molecular evidence for epistatic interactions
(Strobel et al., 2003) – which suggests they are not selectively neutral. In contrast, cognitive
abilities seem to show less VNA (Chipuer, Rovine & Plomin, 1990), a point we consider later.
As summarized in Table 1, genetic variation persists in populations through selective
neutrality only if its phenotypic consequences are (almost) completely unrelated to fitness in
any environment. This genetic variation can be expected to be mainly additive. While it is
possible that this holds for some relatively trivial traits (e.g. gut-packing design), it is highly
implausible for major personality differences, given their pervasive effects on social, sexual,
and familial life.
The evolutionary genetics of personality
13
Can Mutation-Selection Balance Explain Genetic Variance in Personality?
Mutation rates and mutation load. As stated previously, a truly neutral trait has to
show a close-to-null relationship to any fitness component in any environment. All traits that
do not fulfil this very strict requirement are subject to natural selection. As long as the
direction of selection is relatively constant, Fisher’s (1930) fundamental theorem predicts that
the additive genetic variance of the trait will be reduced to the point where one genetic
variant becomes fixed as a universal, species-typical adaptation. The rate of reduction in a
trait’s genetic variance is influenced by two factors with opposing effects: the mutation rate
(which increases genetic variance) and the strength of selection (which decreases genetic
variance). The mutation rate tells us how fast new mutations are introduced into functional
parts of the genome (i.e., protein-coding genes and their regulatory regions). Comparative
molecular genetic studies suggest that humans have a comparatively high mutation rate
(Eyre-Walker & Keightley, 1999), with the best available estimate being an average of about
1.67 new mutations per individual per generation (Keightley & Gaffney, 2003). Given
reasonable assumptions about mutations arising in a Poisson frequency distribution, one can
calculate that the probability of a human being born without any new mutations is slightly
lower than one in five (Keller, in press). Importantly, this estimate includes only non-neutral
mutations (polymorphisms that are visible to selection). As argued above, almost all nonneutral mutations tend to be harmful, and selection is stronger against more harmful
mutations. For example, a mutation that reduces number of surviving offspring by 1% will
persist for an average of ten generations in a large population, passing through the
genotypes of about 100 individuals during that time. A mutation with a weaker 0.1% fitness
reduction (which is still ten times stronger than selective neutrality in humans) will persist for
four generations longer, afflicting about 1,000 individuals (Garcia-Dorado, Caballero & Crow,
2003). Because harmful mutations with dominant effects are an easier target for selection,
only recessive mutations are likely to persist for a longer time (Zhang & Hill, 2005).
The evolutionary genetics of personality
14
It follows that there is a mutation load of older, mildly harmful, and mostly recessive
mutations in any individual at any point in time. This mutation load is mostly inherited from
parents to offspring, but a few new mutations arise in each generation. Thus, each particular
mutation will be eliminated by selection eventually, but at the same time new mutations will
arise. According to very conservative estimates, the average number of mildly harmful
mutations carried by humans is about 500 (Fay, Wyckoff & Wu, 2001; Sunyaev et al., 2001)
and the standard deviation is 22 (or higher, given assortative mating, as we discuss below)
(Keller & Miller, 2006a). This mutation load may account for a substantial portion of genetic
variance in many fitness-related traits – perhaps including personality differences.
Mutational target size. For a long time, Fisher’s fundamental theorem was thought to
imply that traits that affect fitness more strongly should show less VA (Falconer, 1981). In the
early 1990s, however, Price and Schluter (1991) and Houle (1992) showed that the reverse
is true: more fitness-related traits actually tend to have higher VA. The reason that this could
remain unnoticed for more than half a century was that evolutionary geneticists used to
standardize additive genetic variance (VA) by the total phenotypic variance (VP) of the trait,
yielding its narrow-sense heritability (h²):
h² = VA / VP
(2)
Insofar as heritability was taken as a rough proxy for additive genetic variance, this
gives profoundly misleading results, because VP contains both the non-additive genetic (VNA)
and the environmental variance (VE). Even if VA is large, h² can be small when VNA and/or VE
are even larger. Since VE is especially population- and trait-specific, h² is not very informative
for comparing genetic variances. Houle (1992) instead proposed to use the ‘coefficient of
additive genetic variation’ (CVA) for comparisons across traits, populations, and species. It is
defined as:
CVA = sqrt(VA) / M * 100
(3)
The evolutionary genetics of personality
15
or, equivalently,
CVA = sqrt(VP * h²) / M * 100,
(4)
with M being the phenotypic trait mean and 100 a conventional scaling-factor. The CVA thus
standardizes VA by the mean of the trait, whereas h² standardizes VA by its total phenotypic
variance. As long as all traits are measured on a ratio scale and some basic scaling effects
are taken into account (Stirling et al., 2002), CVAs are directly comparable across traits and
species, which does not hold for h²s. For many traits across many species, it turned out that
VA increases with the fitness-relevance of a trait (Houle, 1992; Pomiankowski & Møller, 1994;
Stirling et al., 2002). Because very high residual variances (VNA + VE) often overshadow
substantial VAs, low h² values often fail to reflect this pattern (Rowe & Houle, 1996; Merilä &
Sheldon, 1999; Stirling et al., 2002).
But how could the traits under strongest selection show the highest VAs? The key
seems to be the number of genetic loci that could potentially disrupt the trait by mutating,
which is called the mutational target-size of a trait (Houle, 1998). Since mutations occur with
random probability at any genetic locus, the number of mutations that affect a trait (i.e., its
mutation load) increases linearly with the number of genetic loci that affect the trait. Note that
we are referring to the total number of genetic loci that could potentially affect the trait if they
became polymorphic due to mutation, not the number of loci that are actually polymorphic at
a given point in time (i.e., the QTLs), which are only about 10% of the potential loci
(Pritchard, 2001; Rudan et al., 2003). Fisher’s (1930) fundamental theorem works best for
traits that are affected by only one genetic locus (Price, 1972; Ewens, 1989). The more
genetic loci affect a trait, the greater the probability that any of these loci will be hit by a
mutation, the more mutations will accumulate in the trait, and the harder it will be for
selection to deplete the VA of this trait. Instead of reaching genetic uniformity, non-neutral
traits with large mutational target sizes will therefore be stuck in a balanced state of mutation
and selection.
The evolutionary genetics of personality
16
The trait with the largest mutational target-size is, of course, fitness itself: it is
influenced by all selectively non-neutral parts of the genome (Houle et al., 1994). Fitness
should therefore have a very large CVA, which is in fact the case (Burt, 1995). Similarly, other
traits closely related to fitness (e.g. so-called life history traits, such as longevity or total
offspring number) are usually complex compounds of various heritable traits, leading to high
mutational target sizes. For example, longevity is potentially influenced by disruptions in any
organ system – circulatory, nervous, endocrine, skeletal, etc. – so its mutational target size
includes the mutational target sizes of all these organ systems. Consistent with this, very
high CVAs have been reported for life-history traits in various species (Houle, 1992), including
humans (Miller & Penke, in press; Hughes & Burleson, 2000). In contrast, low CVAs can be
found in genetically simpler traits less related to fitness, such as some morphological traits
(e.g. bristle number in fruit flies or height in humans - Pomiankowski & Møller, 1995; Miller &
Penke, in press).
Insert Figure 1 about here
The watershed model. Cannon and Keller (2005; see also Keller & Miller, 2006a)
introduced the watershed model (Figure 1) as an analogy to illustrate the relation between
genetic variation and the mutational target size of traits. Its basic point is that ‘downstream’
traits, which are closely related to overall fitness, require the adaptive functioning of virtually
the whole organism – the integrated functioning of many subsidiary ‘upstream’ mechanisms behavioural, physiological, and morphological. Just as many small creeks join to become a
stream, and several streams join to become a river, many genetic and neurophysiological
micro-processes (e.g. the regulation of neural migration, axonal myelinzation, and
neurotransmitter levels) might interact to become a specific personality trait. These
personality traits will interact to influence success in survival, socializing, attracting mates,
and raising offspring – which in turn determines overall fitness. The upstream microprocesses, such as the regulation of a particular neurotransmitter, may be influenced by only
a few genes. The broader middle-level processes, such as reactivity to social stress, are
influenced by all genes that affect the corresponding upstream processes. The same holds
The evolutionary genetics of personality
17
true for even broader (i.e., more downstream) domains of organismic functioning – which are
equivalent to broad components of fitness itself (e.g. sexual attractiveness, social status,
foraging efficiency) – these depend on all of the genes that affect all of their upstream
processes. A similar argument holds for environmental influences, which, when affecting
upstream processes, accumulate in downstream traits. But because selection is much less
effective in reducing VE, the VE of fitness components tends to be large, which reduces their
heritability. Merilä and Sheldon (1999) argued that VNA is as robust against selection as VE,
which would imply a high Dα for traits under mutation-selection balance. However, more
recent evidence questions the robustness of VNA to selection in downstream traits (Stirling et
al., 2002). The exact expected size of Dα for traits under mutation-selection balance must
thus be regarded an unresolved issue, though it is likely in the medium range.
Developmental stability and the f-factor. As an addition to the watershed model,
developmental stability theory (Polak, 2003) explains how mutations that are spread across
the genome influence fitness. It argues that organisms often fail to develop according to the
evolved blueprint in their genome, since either the blueprint itself or the relevant
environmental factors are disrupted. In such a case, the evolved fit between genome and
environment is disrupted. Whereas the genomic blueprint is disrupted by mutations, the
organism’s developmental environment can be disrupted by factors such as pathogens and
toxins. From a fitness perspective, the exact combination of disruptive factors doesn’t matter:
what counts is the total reduction in phenotypic functionality due to developmental instability.
Similarly, only the total mutational damage in the genome is what counts for natural
selection. Which genetic sequences the mutations disrupt are largely unimportant – and
likely different for each human being.
An established measure of developmental stability is the bilateral symmetry of body
parts that show perfect symmetry at the average population level (e.g. ankle breadth or ear
length), usually aggregated across many body parts. Even though this only taps into
morphological developmental stability, body symmetry shows relations to all kinds of fitness
components in various species (Møller, 1997), including humans (Gangestad & Yeo, 1997;
The evolutionary genetics of personality
18
Gangestad & Simpson, 2000). One well-replicated correlate of body symmetry is general
intelligence (Prokosch et al., 2005; Luxen & Buunk, 2006; Bates, 2007). Thus, some genetic
and environmental disruptions can apparently impair both cognitive and morphological
development. The watershed metaphor breaks down a bit at this point, because it fails to
reflect the fact that most mutations are pleiotropic in their effects (Marcus, 2004): each
mutation will tend to disrupt several downstream traits. Those harmful effects will be
positively intercorrelated in the affected downstream traits (not because the effects are
positive, but because they are consistently negative). Therefore, pleiotropic mutations should
lead to a ‘positive manifold’ of intercorrelations among the efficiencies of mid-level processes
and of fitness components. In addition, intercorrelations between various processes may
arise through developmental interdependence (van der Maas et al., 2006). According to
Miller (2000c), this should allow the extraction of a ‘general fitness factor’ or ‘f-factor’ that
reflects (inverse) overall mutation load. Just as the g-factor of general intelligence (Jensen,
1998) is at the top of a multi-level hierarchy of intercorrelated cognitive abilities, f is at the top
of a similar hierarchy of genetically intercorrelated upstream traits and processes. In fact,
Miller and colleagues (2000c; Prokosch et al., 2005) argued that g is an important subfactor
of f, reflecting the integrative functioning of the cognitive system. The VA of g may therefore
reflect the aggregate harmful effects of mutations at any of the thousands of genetic loci that
affect our brain development and functioning, each of which decreases our cognitive abilities
a tiny bit.
Further predictions. Every trait under mutation-selection balance has to be a
downstream trait, with mutations occurring randomly across all of the loci that contribute to its
mutational target size. It is very unlikely that any of these harmful mutations will ever reach
an intermediate prevalence rate in the face of selection working against it (Turelli & Barton,
2004). The mutations that cause the VA of more complex downstream traits will thus be
numerous, but individually rare, evolutionarily transient, and phenotypically mild in their
effects. As a consequence, they will be extremely hard to detect using standard molecular
genetic methods (linkage and association studies), and they will be very unlikely to replicate
The evolutionary genetics of personality
19
across populations (because different evolutionarily transient mutations tend to affect
different populations). Furthermore, since the sheer number of involved loci will impede
selection’s ability to deplete VA, the magnitude of Dα for downstream traits will likely be in the
medium range (Stirling et al., 2002). These predictions (see Table 1) are consistent with what
is currently known about the genetic structure of g (Plomin, Kennedy & Craig, 2006; Plomin &
Spinath, 2004). Enormous efforts to identify single genes of major effect underlying
intelligence led to meagre success at best, and to the conclusion that a huge number of
pleiotropic polymorphisms must be responsible for its genetic variation (Kovas & Plomin,
2006). The situation is different for personality traits, however, since good candidates for
underlying polymorphisms have been identified (Ebstein, 2006), and most of these have
intermediate prevalence rates (Kidd, 2006). In addition, the amount of VNA found in
personality traits is often as high as the VA component (Eaves et al., 1998; Keller et al.,
2005), indicating a large Dα of .50 or higher. These characteristics of personality traits cannot
be explained by mutation-selection balance.
Since traits with a large mutational target size tend to be most affected by mutations
that are both rare and recessive, the probability that two copies of the same mutation come
together in a single individual and unleash their full deleterious potential is much higher when
both parents are genetically related. This is called inbreeding depression. Its counterpart is
called heterosis or outbreeding elevation, and occurs when pairings of recessive, deleterious
mutations are broken up by sexual recombination in offspring of highly unrelated parents
(e.g. parents from different ethnic groups). Due to the predicted genetic structure of traits
under mutation-selection balance, we can expect them to show both inbreeding depression
and heterosis effects (DeRose & Roff, 1999; Lynch & Walsh, 1998). Such evidence exists for
intelligence (reviewed in Jensen, 1998), but is, to the best of our knowledge, absent for
personality traits. For example, the offspring of cousin marriages tend to be less intelligent,
but we do not know of any evidence that they tend to be more or less extraverted,
conscientious, or agreeable than average.
The evolutionary genetics of personality
20
Finally, the typically harmful effects of mutations lead to a clear prediction about the
social perception of their phenotypic effects. Since a high mutation load disrupts an
organism’s functional integrity and ultimately fitness, it should lead to a less favourable social
evaluation by those who are looking for a good sexual partner, friend, or ally. The mating
context is most important here, because about half of a sexual partner’s mutation load will be
passed along to one’s offspring (Keller, in press). Indeed, virtually all modern evolutionary
theories of mate choice argue that any phenotypic trait that reliably signals that a potential
mate has a low mutation load will be sexually attractive (Keller, in press; Kokko, Brooks,
Jennions & Morley, 2003; Miller, 2000b, c). In an influential paper, Rowe and Houle (1996)
argued that sexual selection would drive the evolution of any sexually attractive trait towards
higher reliability by making its expression more condition-dependent, that is more dependent
upon (and revealing of) the overall phenotypic condition (e.g. health, vigour) of the organism.
Condition is a trait with very large mutational target size, near the downstream end of the
watershed model (Figure 1), and very closely related to fitness (Tomkins, Radwan, Kotiaho &
Tregenza, 2004). A condition-dependent trait is thus affected by larger parts of the genome –
it will actually ‘move downstream’, insofar as it becomes sensitive to the efficiency of a larger
number of upstream processes. This can explain why, across species, morphological traits
that are preferred in mate choice (e.g. the plumage of finches) tend to have much higher CVA
than morphological traits that are irrelevant for mate choice (e.g. bristle number in fruitflies)
(Pomiankowski & Møller, 1995), and almost as high as extreme downstream traits such as
longevity and fertility.
Since traits that reliably reveal genetic quality (low mutation load) and general
phenotypic condition tend to be highly variable within each sex and highly attractive to the
other sex, mating markets in socially monogamous species (such as humans) tend to be
competitive. Each individual tries to attract the highest-quality mate who will reciprocate his
or her interest. Given a period of mutual search in such a competitive mating market, socially
monogamous couples tend to form that are closely matched on the average attractiveness
level of their sexually attractive traits (Penke, Todd, Lenton & Fasolo, in press). This
The evolutionary genetics of personality
21
phenomenon, called assortative mating (Vandenberg, 1972), is a typical population-level
outcome for traits that are under mutation-selection balance, but it is much less likely for
traits that are less related to fitness. Mate preferences for higher intelligence, and assortative
mating with respect to intelligence, are well-established phenomena in humans, as is the
condition-dependent expression of intelligence (Miller, 2000c; Miller & Penke, in press). In
contrast, mate preferences for personality traits tend to be modest in size and variable
across individuals (Figueredo, Sefcek & Jones, 2006). In addition, there is almost no
assortative mating for personality traits (e.g. Vandenberg, 1972; Lykken & Tellegen, 1993;
Eaves et al., 1999). Thus, mate preferences for personality traits show quite a different
pattern than mate preferences for universally sought traits, such as intelligence, mental
health, and physical attractiveness– which are all presumably condition-dependent and
under mutation-selection balance.1
To summarize, mutation-selection balance is a very plausible mechanism for
maintaining genetic variation in traits that reflect the overall functional integrity of the
organism, including general intelligence and general health. This is reflected in the following
features: high additive genetic variation, an elusive molecular genetic basis, conditiondependence, inbreeding and outbreeding effects, strong mate preferences, and assortative
mating (see Table 1). Personality traits do not match these features nearly as well,
suggesting that mutation-selection balance may not account for much genetic variance in
personality traits.
Can Balancing Selection Explain Genetic Variance in Personality Traits?
In both selective neutrality and mutation-selection balance, genetic variation is
maintained because selection is unable to deplete it – either because the variation is
selectively neutral, or because too much new variation is continually reintroduced. A quite
different mechanism is the maintenance of genetic variation by selection itself. This only
works if the selective forces that act on a trait are balanced, which occurs when both
The evolutionary genetics of personality
22
extremes of the same trait dimension are favoured by selection to the same degree under
different conditions. Such balancing selection can happen in a variety of ways.
Variants of Balancing Selection. One form of balancing selection is overdominance
(also called heterozygous advantage), which occurs when individuals with different alleles at
the same genetic locus have a higher fitness than individuals with two identical copies. Other
Sickle-cell anaemia is a famous textbook case of overdominance, but other examples have
rarely been found in nature (Endler, 1986) and or in animal experiments (Maynard Smith,
1998). Also, it is now widely believed that overdominance is evolutionary unstable and thus
an unlikely candidate for maintaining genetic variation, especially in the long-term (Roff,
1997; Keller & Miller, 2006a; Bürger, 2000).
Another form of balancing selection is antagonistic pleiotropy, which occurs when
polymorphisms have a positive effect on one fitness-related trait and a negative effect on
another (Roff, 1997; Hendrick, 1999). A special case is sexually antagonistic co-evolution,
where genetic variants are under opposing selection pressures in men and women (Rice &
Chippindale, 2001). Since selection will usually fix the polymorphism with the least total
fitness cost, antagonistic pleiotropy could only maintain genetic variation if the fitness costs
of all alleles at such a locus are exactly equal (averaged across environments). In addition,
all heterozygous allele combinations have to provide all phenotypic fitness benefits that
would be provided by both corresponding homozygous combinations (‘reversal of
dominance’, Curtisinger, Service & Prout, 1994; Hendrick, 1999). Furthermore, independent
of the number of genetic loci that affect a quantitative trait, antagonistic pleiotropy can
maintain genetic variation only at one genetic locus (or two in the case of sexually
antagonistic co-evolution) per trait (Turelli & Barton, 2004). Due to these highly restrictive
conditions, it is very unlikely that antagonistic pleiotropy plays a major role in maintaining
genetic variation (Hendrick, 1999) – although the special case of sexually antagonistic coevolution might contribute to sex differences in personality and some within-sex personality
variation (Keller & Miller, 2006b).
The evolutionary genetics of personality
23
A more likely variant of balancing selection is environmental heterogeneity. When a
trait’s effect on fitness varies across space or time, significant genetic variation can be
maintained in populations (Roff, 1997), even in quantitative traits (Bürger, 2000; Turelli &
Barton, 2004). A necessary requirement for this to happen is that spatial or temporal
fluctuations in selection pressures must occur such that the trait’s net fitness effects are
nearly neutral when averaged across all relevant spatiotemporal environments. It is not
enough for a trait to be neutral in some environments or during some periods, because
selection is very efficient at favouring polymorphisms with higher average fitness outcomes
across all relevant environments. Only a fully balanced effect of different alleles across space
and time will work to maintain genetic variation.
A related type of balancing selection is called frequency-dependent selection. In this
case, the spatiotemporal fluctuations in selection pressures usually occur in the social
environment of the species, rather than the external physical environment. Frequencydependent selection can only maintain genetic variations if it is negative, favouring traits as
long as they are rare in frequency (Maynard Smith, 1998). (Positive frequency-dependence
will drive polymorphisms to fixation through a runaway, winner-take-all effect.) The ‘social
environment’ is used in a very broad sense here, and can include the ratio of cooperative
partners to cheaters (Mealey, 1995), the ratio of males to females (Fisher, 1930), the
distribution of intra- and interspecific competitors for limited resources in ecological niches, or
even parasite-host relationships (which occurs when viruses, bacteria or other pathogenes
are best adapted to exploit the most common host phenotypes - Garrigan & Hedrick, 2003).
Mathematical models have shown that negative frequency-dependent selection in any of
these ways is a viable way to maintain genetic variance (Bürger, 2005; Schneider, 2006).
Thus, environmental heterogeneity and negative frequency-dependent selection are
good candidates for maintaining genetic variance by balancing selection, whereas
overdominance and antagonistic pleiotropy can work only in rare cases that meet very
restrictive conditions. The bottom line is that balancing selection requires a set of varying
selection pressures that favour different phenotypes under different conditions. These
The evolutionary genetics of personality
24
fluctuating selection pressures must be stronger than any other unidirectional selection
pressures on the same trait that consistently favour a certain optimal trait level in every
environment (Turelli & Barton, 2004). If this condition is met, balancing selection leads to two
or more different phenotypes (or a continuum of phenotypes) with identical average fitness
across environments. Since these phenotypes cannot be further optimized by selection, they
are called evolutionary stable strategies (ESSs) (Maynard Smith, 1982).
Predictions. Balancing selection leads to some distinctive genetic patterns.
Reoccurring periods of selection in different directions tend to deplete the VA of affected traits
and result in higher Dα than found for selectively neutral traits (Roff, 1997). Dα will also be
higher for traits under balancing selection than for traits under mutation-selection balance,
since the former maintains polymorphisms at fewer genetic loci than the latter (Kopp &
Hermisson, 2006), and selection is more effective in depleting the VA from fewer genetic loci
(van Oers et al., 2005; Stirling et al., 2002). Furthermore, balancing selection can maintain
alleles in a population at intermediate prevalences, while mutation-selection balance cannot
(Turelli & Barton, 2004). These characteristics (as summarized in Table 1) make balancing
selection a likely candidate for maintaining genetic variation in personality traits, although it is
unlikely to explain persistent genetic variance in cognitive abilities.
Balancing Selection and Personality Traits. When Tooby and Cosmides (1990)
argued that heritable personality differences are basically evolutionary noise, they suggested
that parasite-host co-evolution (Garrigan & Hedrick, 2003), a form of negative frequencydependent selection, might explain the striking amount of evolutionary ‘noise’ in human
behavioural traits better than selective neutrality. Nonetheless, the central message was the
same for both evolutionary processes: since the heritable aspects of personality are random
by-products of functionally superficial biochemical differences that exist - at best - to prevent
our lives from parasites, studying personality differences from an evolutionary perspective is
a big waste of time. However, as argued above, there is strong evidence that personality
differences have direct effects on fitness. In addition, Keller and Miller (2006a) noted that, for
parasite-host co-evolution to explain personality variation as a by-product, there would have
The evolutionary genetics of personality
25
to be a very high degree of overlap between genetic loci that affect immune system function
and genetic loci that affect personality differences – which seems unlikely.
MacDonald (1995, 1998) made an important step away from Tooby and Cosmides’
‘neutral personality assumption’ by proposing that five independent behavioural systems
under balancing selection explain the dimensions of the Five Factor Model of personality
(FFM). While he regarded both extremes of each dimension as maladaptive, with stabilizing
selection working against them, he assumed that the relatively broad middle range of each
personality dimension reflects equally viable behavioural strategies (i.e., ESSs). MacDonald
(1998) also argued that the viability of these strategies should vary across environmental
niches. Following MacDonald (1995, 1998), Nettle (2006a) developed more specific
hypotheses about the potential fitness costs and benefits associated with each of the FFM
dimensions. If these evolutionary cost-benefit trade-offs were exactly the same in every
environment, they could maintain genetic variance only through antagonistic pleiotropy,
which tends to be evolutionary unstable. However, if the relevant selection pressures
fluctuate across time or space, favouring different optima on the cost-benefit curves, they
could maintain the range of viable personality trait levels. For example, Nettle (2006a)
argued that high extraversion yields fitness benefits by promoting mating success, social
alliance formation, and environmental exploration, but at the cost of increased physical risks
and decreased romantic relationship stability. When environments are physically riskier to
oneself and one’s offspring (who benefit from relationship stability), high extraversion may be
a net fitness cost; but when conditions are safer, high extraversion may yield a net fitness
benefit. Environmental fluctuations would thus maintain genetic variation in extraversion.
The challenge in any such balancing selection argument is to identify the specific
costs and benefits relevant to each personality trait across different environments. Originally,
Nettle (2005) also hypothesized that extraverts might conserve energy by investing less
parental effort in offspring, but failed to find supportive evidence. In fact, Nettle’s list of
extraversion costs and benefits might still be too long, with some proving to be fitnessirrelevant by-products. On the other hand, these are only some of the plausible costs and
The evolutionary genetics of personality
26
benefits. Different ones can be suggested for this and other personality traits (Denissen &
Penke, 2006). Even if balancing selection proves a good general account of heritable
personality traits, much more research would be needed to identify each personality trait’s
relevant fitness costs and benefits across different environments.
Environmental Niches for Personality Traits. Recently, Ciani and colleagues (Ciani,
Veronese, Capiluppi & Sartori, in press) reported an interesting natural experiment that
indirectly supports a role for balancing selection by environmental heterogeneity in sustaining
the genetic variance of personality traits. They studied average personality differences on the
FFM dimensions of Italian coast-dwellers compared to Italians living off the coast on three
small island groups. After matching populations for cultural, historical and linguistic
background, and controlling for age, sex and education, they found that individuals from
families that have lived on small islands for at least 20 generations were lower in
extraversion and openness to experience than both mainlanders and more recent immigrants
to the island. This pattern makes cultural or developmental explanations for the population
differences unlikely - it suggests change on the genetic level. Even though individual fitness
consequences of these traits were not measured directly, the apparent recent evolution of
genetic differences between populations in these two traits suggests that the fitness payoffs
of these two personality traits were historically distinct in these different environments.
In non-human species, recent studies suggest that environmental heterogeneity does
impose varying selection pressures on personality traits. Dingenmanse, Both, Drent and
Tinbergen (2004) could directly measure the fitness payoffs of personality differences (on a
carefully assessed shyness-boldness dimension) in the great tit (parus major), which varied
with food availability across breeding seasons. Similar evidence of environmental
heterogeneity favouring personalities exists for some other species (reviewed in
Dingenmanse & Réale, 2005).
More direct evidence for the importance of environmental heterogeneity in the
evolutionary genetics of human personality comes from studies of the global distribution of
polymorphisms at the DRD4 locus. This gene regulates dopamine receptors in the brain and
The evolutionary genetics of personality
27
has been associated with personality traits such as novelty seeking and extraversion
(Ebstein, 2006). The prevalences of different DRD4 alleles differ dramatically across world
regions. The evolutionarily newer 7R allele, which is more common in risk-prone, responseready, extraverted novelty seekers, is much more prevalent in European and American
populations than in Asian populations (Chang et al., 1996). This allele appears to be
favoured by selection (1) when benefits can be gained from migrating to new environments
(Chen et al., 1999; Ding et al., 2002), and (2) under resource-rich environmental conditions
(Wang et al., 2004). Referring to these findings, Harpending and Cochran (2002) noted that
under conditions of environmental harshness and resource scarcity (as is common in huntergatherer societies), intensive cooperation, strong family ties, stable pair bonds, and
biparental investment are necessary for survival and successful reproduction. These
ancestrally typical conditions would maintain the more risk-averse, ancestral form of the
DRD4 gene. But under more luxuriant environmental conditions, when children can survive
without so much paternal support (as in most agricultural and modern societies), the more
risk-seeking 7R allele should be favoured by selection, as it leads to a personality more
prone to sexual promiscuity and intrasexual competition (see also Gangestad & Simpson,
2000; Schmitt, 2005).
Arguments for frequency-dependent selection. The role of competition demands
some more attention here. Competition, whether for mates, food, or other limited resources,
is often a zero-sum game: The winner gains a benefit, but the loser usually pays a cost, at
least in the form of wasted time and effort. As the competition within a niche becomes more
intense, selection may eventually favour less competitive individuals who refrain from
seeking these benefits to avoid the associated costs. This is the logic of the so-called ‘hawkdove game’, the classic example of negative frequency-dependent selection (Maynard Smith,
1982). In fact, some evolutionary geneticists have argued that most environmental niches are
actually social in nature, because the fluctuating selection regimes caused by environmental
heterogeneity are almost always mediated by within-species competition that often takes the
form of negative frequency-dependent selection (Bürger, 2005; Kassen, 2002). It is
The evolutionary genetics of personality
28
interesting in this regard that personality differences have been found almost exclusively in
social species (Figueredo et al., 2005a) and that they tend to have stronger effects on fitness
over social than non-social paths in most species (Smith & Blumstein, 2007). Personality
appears to be fundamentally social, perhaps reflecting the diversity of social and sexual
strategies that can prosper in socially variegated groups that confront fluctuating,
heterogeneous environments. This might be especially true for human personality after our
species achieved ‘ecological dominance’ (i.e. reliable mastery of food acquisition and
protection from predators and other hazards), which somewhat buffered our ancestors from
spatiotemporal variation in the non-social environment (Alexander, 1989). Explicit arguments
that negative frequency-dependent selection could maintain genetic variance in specific
personality traits have been proposed by Gangestad & Simpson (1990) for female
sociosexuality (i.e., promiscuity) and by Mealey (1995) for psychopathy.
Another application of negative frequency-dependent selection to explain personality
has been proposed by Rushton (1985) and extended by Figueredo et al. (2005a, b). They
argue that virtually all human individual differences, including broad personality factors,
intelligence, attachment styles, reproductive strategies, growth, longevity, and fecundity, may
reflect a single underlying ‘life-history’ dimension of variation in the organism’s allocation of
investment in growth vs. survival vs. reproduction across the life-course. Drawing a parallel
to a similar, well-established dimension of between-species differences in evolutionary
ecology, they suggest that this life-history dimension is maintained by negative frequencydependent selection within and across human groups. A fortuitous side-effect is that such
variation reduces within-group and between-group competition by allowing individuals and
groups to fill different socio-environmental niches. Figueredo et al. (2005a, b) hypothesized
that if a broad set of physical and psychology traits (e.g. intelligence, personality traits,
sociosexuality, longevity) are subject to hierarchical factor analysis, a superordinate ‘K-factor’
will emerge that reflects variation on this life history dimension (note that this hypothesized Kfactor is distinct from the f-factor discussed above).
The evolutionary genetics of personality
29
A critical point from an evolutionary genetic perspective is that frequency-dependent
selection (like any form of balancing selection) is only able to maintain polymorphisms at a
few major loci (Turelli & Barton, 2004; Kopp & Hermisson, 2006). As a consequence,
frequency-dependent selection on the K-factor would only be possible if a few
polymorphisms would function as ‘switches’ that could simultaneously alter the development
and expression of all those many traits the K-factor aims to explain, including some of the
most important emergent traits at the downstream end of the watershed model (Figure 1),
such as longevity, growth, intelligence, and fecundity. As long as there is no evidence that
these ‘polymorphisms for almost everything’ exist, future research on life history variation
should distinguish more carefully between (1) mutation-selection balance for downstream
traits like longevity, growth, intelligence, and fecundity, (2) the condition- and environmentdependent adjustment of reproductive strategies (Gangestad & Simpson, 2000; Penke &
Denissen, 2007), and (3) balancing selection for various independent personality traits at a
more upstream level of genetic complexity.
To summarize, balancing selection by environmental heterogeneity, often mediated
by negative frequency-dependent selection, seems the most plausible mechanism for
maintaining genetic variation in personality traits. In contrast, balancing selection is
implausible for maintaining genetic variation in downstream fitness-related traits, such as
intelligence.
The Role of the Environment in Evolutionary Genetics
Evolutionary adaptationism is often misunderstood as overemphasizing genetic
influences and neglecting environmental influences on behaviour. In fact, the opposite is
generally true: evolutionary theory is fundamentally environmentalistic (Crawford &
Anderson, 1989), because it is about the adaptive fit of an organism to its environment – a
GxE interaction.
Phenotypic plasticity. One form of this interaction – selection - has already been
discussed. Selection acts only upon the complete phenotype, which is at the most
The evolutionary genetics of personality
30
downstream end of the watershed model (Figure 1), at the level of overall fitness. But GxE
interactions take place all the way upstream, up to the molecular level, where transcribed
genes can only produce specific proteins if the required amino acids are present (ultimately a
nutritional issue). From this perspective, it is hardly surprising that identical genotypes can
produce very distinct phenotypes. This phenomenon is called phenotypic plasticity, and it is
probably ubiquitous in nature (West-Eberhard, 2003). The environment thus has two distinct
roles in evolutionary genetics: It interacts with the genotype in the ontological development of
the phenotype, and then, as a selective regime, determines the phenotype’s fitness and
decides its fate.
Ideally, organisms would fare best if they could fit themselves perfectly and instantly
to the environmental demands in every situation – morphologically, physiologically and
behaviourally. Of course, developmental constraints render such an unlimited degree of
phenotypic plasticity implausible for physical traits (e.g., no drowning mammal can suddenly
develop gills, no matter how advantageous such a transformation would be). In contrast,
unlimited behavioural plasticity has been an attractive scientific vision for a long time, both in
psychology (i.e., radical behaviourism) and biology (i.e., traditional behavioural ecology;
Krebs & Davies, 1997). But even in the case of behaviour, unlimited plasticity is impossible to
achieve adaptively, because the environment does not reliably signal the likely fitness
payoffs of all possible behavioural strategies (see Miller, in press). In a complex world,
environmental cues that can guide adaptive behaviour are inherently noisy, often
contradictory, and unpredictably variable (Brunswick, 1956; Gigerenzer, Todd & the ABC
Research Group, 1999). The unreliability of environmental cues means that any behavioural
plasticity based on trial-and-error learning must take time, because it must depend upon a
decent sample of action-payoff pairings. Thus, given the complexities of real-world
environments, organisms cannot instantly discern and implement the optimal behavioural
strategy, so fitness-maximizing by unlimited behavioural plasticity is an impossible ideal.
Universal constraints on phenotypic plasticity. Fortunately, evolution constrains
behavioural plasticity in adaptive directions, just as it constrains physical development. As
The evolutionary genetics of personality
31
long as environmental features are sufficiently stable and fitness-relevant (e.g. women get
pregnant but men don’t, rotten food is toxic, children demand more care and protection than
adults), natural selection will fixate psychological mechanisms such as emotions,
preferences, and learning preparednesses that adaptively bias our reactions to the
environment over ontogenetic development. This relieves us from the impossible task of
learning our most basic behavioural dispositions de novo every generation (Tooby,
Cosmides & Barrett, 2005; Barrett, 2006; Figueredo et al., 2006). These kinds of GxE
interactions – interactions between inherited psychological adaptations and ancestral
adaptive challenges - are the central subject of adaptationistic evolutionary psychology.
Cervone (2000) argued that they also constitute interesting building blocks for personality
theories. However, adaptationistic evolutionary psychology deals principally with interactions
between the universal genetic make-up of our species and fitness-relevant aspects of the
environment that reoccurred over evolutionary time. Such interactions might explain the nongenetic variation in some personality domains (e.g. attachment styles - Buss & Greiling,
1999), but are largely uninformative about heritable personality differences.
Individual constraints on phenotypic plasticity. When selection cannot deplete all
genetic variation (for any of those reasons discussed above), different genotypes persist
simultaneously in the population. Genotypes might differ in their response to the
environment, leading to the statistical effect that behaviour geneticists refer to as a GxE
interaction (Moffitt, Caspi & Rutter, 2006). In humans, such interactions have been found, for
example, between the MAOA polymorphism and childhood maltreatment in the development
of conduct behaviour (Caspi et al., 2002), and between the 5-HTT polymorphism and
stressful life events in the development of depressiveness (Caspi et al., 2003). By
systematically varying both the genotypes and the environments, evolutionary geneticists
studying non-human species can determine a typical response function for each individual
genotype, a so-called reaction norm (Via et al., 1995) (see Figure 2). While a GxE interaction
is a population statistic, an individual reaction norm can be regarded as a characteristic of an
individual genotype (Pigliucci, 2005). Reaction norms were originally used to study the
The evolutionary genetics of personality
32
developmental plasticity of morphological or life-history traits, but when behavioural
ecologists realized the systematic limits of behavioural flexibility, they began to view heritable
response styles – known to psychologists as personality traits – as behavioural reaction
norms. (Sih et al., 2004; van Oers et al., 2005).
Insert Figure 2 about here
While behavioural ecologists discovered animal personality only recently (Sih et al.,
2004), their immediate equation of personality traits with individual reaction norms helped
them to circumvent the ‘person-situation debate’ in personality psychology (Mischel, 2004).
Instead of looking for personalities that reliably predict behaviour across all possible
situations, or situations that reliably predict behaviour across all possible personalities,
behavioural ecologists quickly adopted a reaction-norm view of personality that neatly
resembles the personality signatures view of Mischel and Shoda (1995). Personality
signatures describe stable patterns of contingent (if-then) relationships between
personalities, situations, and behaviours – just as reaction norms describe stable
contingencies between genotypes, environments, and phenotypic outcomes. These personsituation contingency profiles turn out to show reasonable consistency (Mischel & Shoda,
1995; Borkenau, Riemann, Spinath & Angleitner, 2006), but it is a different type of
consistency than the well-known rank-order stabilities of personality traits across situations
(Mischel, 2004). However, unlike individual reaction norms, personality signatures describe
environment-behaviour functions for persons, not for genotypes. Although Mischel and
Shoda (1995) acknowledge the possibility that genes influence personality signatures, their
Cognitive-Affective Personality Systems model emphasises the importance of learned
beliefs, appraisals, expectancies, and goals, organized in cognitive-affective units. However,
personality signatures show substantial heritabilities (Borkenau et al., 2006), so these
cognitive-affective units are apparently influenced by genetic variation, and a genotypeoriented reaction-norm view may be appropriate.
To describe an individual reaction norm does not require a mechanistic model of the
psychological processes that mediate between environmental contingencies and behaviours.
The evolutionary genetics of personality
33
Reaction norms simply relate dimensional variations in genotypes and environments to
variations in behavioural outcomes. Thus, the shapes of individual reaction norms are what
can be equated with personality traits (van Oers et al., 2005). While reaction norm shapes
can be simple (e.g. linear) when relating polymorphisms at a single gene locus to the
environment (as for example in Caspi et al., 2003), they can become more complex when
polygenic genotypes (as in the case of personality traits) are related to the environment (de
Jong, 1990). Furthermore, while the studies by Caspi et al. (2002, 2003) provide examples of
reaction norms in personality development (i.e., GxE interactions during childhood predict
adolescent personality), the concept of individual reaction norms is not limited to a
developmental time frame. Reaction norms can also describe GxE interactions in the
production of ongoing behaviour, analogous to Mischel and Shoda’s (1995) personality
signatures.
Note that reaction norms can be determined for any phenotypic trait, including
cognitive abilities. However, we believe that reaction norms are much more informative for
personality traits than for cognitive abilities. Reaction norms provide an elegant tool to
disentangle the twofold role of the environment for personality traits as both a source of
phenotypic plasticity within a generation and of fluctuating selection pressures across
generations. This more nuanced view of environmental influences on behaviour is
unnecessary for fitness components such as cognitive abilities that are more likely under
mutation-selection balance, in which case selection pressures push traits in roughly the
same direction (minimum genetic mutation load, maximum phenotypic efficiency) across all
kinds of environments. In addition, the phenotypic plasticity of general intelligence apparently
reflects simple condition-dependency, as g declines with adverse environmental influences
(e.g. starvation, dehydration, sickness) that decrease general condition (see Miller & Penke,
in press). Since the genetic variation in g accounts for almost all genetic variation in cognitive
abilities (Plomin & Spinath, 2004), the reaction-norm view seems less helpful for cognitive
abilities than for personality traits.
The evolutionary genetics of personality
34
Individual Reaction Norms and the Hierarchical Structure of Personality Traits
The hypothesized existence of complex individual reaction norms has an interesting
implication for the hierarchical structure of personality traits. We illustrate this with an
example modified from van Oers et al. (2005) (Figure 2): Let two personality traits (say,
depressiveness and anxiousness) be described by reaction norms to a continuum of
environmental stress. For depressiveness, we assume the simple reaction norm found by
Caspi et al. (2003) (Figure 2a): Genotype A shows high depressiveness in highly stressful
environments (i.e., point Z), medium depressiveness in the less stressful environment Y, and
no depressiveness in the calm environment X. Genotype B shows the same reaction on a
lower level (i.e., B’s individual reaction norm has a smaller slope), while C is resilient in all
environments. Let us now assume a hypothetical, more complex reaction norm for
anxiousness based on the same three genotypes and environments (Figure 2b). In
environment Z, the rank order of the anxious reactions is the same as for depressive
reactions for the three genotypes (A > B > C), implying a positive genetic correlation between
the two traits in this environment. (Note that the reaction norm model assumes that all
relevant environmental influences are captured either in the environmental dimension or in
confidence intervals around the reaction norm functions, so that we can speak of genetic
correlations here.) The critical effect of complex reaction norms is revealed at the other two
points of the environmental dimension: In environment Y, genotypes A and C react with an
identical degree of anxiety, and genotype B reacts only slightly more strongly. The genetic
correlation between anxiety and depressiveness in this environment would therefore be close
to zero. Finally, in environment X, the rank order of the anxious reactions for the three
genotypes is the inverse of their rank order for depressive reactions in the same
environment, leading to an apparent negative genetic correlation. In this purely hypothetical
example, subsuming both traits in a higher order factor (here neuroticism) would not be
warranted, since their relationship is highly context-dependent. More generally, delineating
hierarchical personality structures would be impeded by sign changes in the genetic
correlations among personality traits measured across environments. Therefore, van Oers et
The evolutionary genetics of personality
35
al. (2005) regard the absence of sign changes in genetic correlations of related facet traits
across environments as a necessary condition for the existence of superordinate personality
domains. This leads us to specific requirements concerning how personality-related genes
must affect multiple personality traits
Structural pleiotropy. Except for some rare and evolutionary unstable cases (called
linkage disequilibria), genetic correlations are always caused by pleiotropy, the effect of
polymorphisms on multiple traits (Roff, 1997). Pleiotropy has been shown for the hierarchical
structure of the FFM in twin studies (Yamagata et al., 2006; Jang et al., 1998, 2002; McCrae
et al., 2001). But as in our hypothetical example, pleiotropy in itself does not prevent sign
changes in genetic correlations between traits across environments. Sign changes can only
be prevented by functional, physiological, or developmental links between the effects of
polymorphisms on one trait and their effects on another trait. Such a condition, called
structural pleiotropy, poses a developmental constraint on the independent phenotypic
expression of both traits in all environments (de Jong, 1990). To be sure, structural pleiotropy
does not mean that complex reaction norms, such as those depicted in Figure 2b, are
theoretically implausible. Instead, the central point is that, for two traits to be facets of the
same higher-order factor, the rank order of the phenotypic effects produced by different
genotypes must not reverse across environments. The traits in Figures 2a and 2b cannot
belong to the same higher-order factors, but both can, together with other traits, belong to
different factors.
An implication of structural pleiotropy is the existence of underlying
neurogenetic mechanisms (e.g. neurotransmitter or endocrinological systems) that are
shared by all facets of a higher-order trait. An advantage of viewing personality traits as
individual reaction norms is that these mechanisms, which should be closely linked to the
genotype, can be explicitly separated from the environmental factors with which they interact.
In this way, individual reaction norms come much closer to the original personality trait
definition by Allport (1937) as “psychophysical systems that determine [an individual’s]
unique adjustment to his environments” (p. 48), than to the purely descriptive, empirically
The evolutionary genetics of personality
36
derived factors that are normally posited in personality psychology, and they also avoid the
often-criticized circularity of the definition of traits as aggregated instances of behaviour,
which are then used to predict…behaviour (see Denissen & Penke, 2006).
A developmental perspective. If broad personality domains exist because of shared
underlying mechanisms (i.e., because structural pleiotropy preserves the sign of genetic
correlations between traits across environments), then personality structure likely develops
top-down, from these mechanisms to higher-order personality domains (e.g. neuroticism) to
lower-order personality facets (e.g. anxiousness, depressiveness). Over the lifespan, these
mechanisms might modulate the cognitive and affective experiences that individuals acquire
through interacting with their environments Thereby, they might act as forms of ‘prepared
learning’ (Figueredo et al., 2006) for the acquisition of the cognitive-affective units
emphasized by Mischel and Shoda (1995), and as ‘experience-producing drives’ (Bouchard
et al., 1996) that motivate active niche selection (Denissen & Penke, 2006). These shared
mechanisms would be the ties that bind different facet traits within broader personality
domains. Together with the influence of unique genetic variation on the level of lower-order
traits (Jang et al., 1998, 2002), this would result in the hierarchical structure of personality
traits, down to the level of idiosyncratic habits and behavioural patterns.
The dimensionality of personality. Note that this theoretical argument makes no
commitment to any particular number of highest-order mechanisms or their interactions. The
prominence of the FFM led evolutionary psychologists (MacDonald, 1995, 1998; Nettle,
2006a), including us (Denissen & Penke, 2006), to hypothesize selection regimes at this
hierarchical level. However, some of the FFM dimensions may still share some common
mechanisms that render them not entirely orthogonal (Jang et al., in press). For example,
Jang et al. (2001) showed a significant amount of genetic overlap between the domains of
neuroticism and agreeableness, which was partly explained by the 5-HTTLPR polymorphism.
It is also possible that several neurogenetic mechanisms interact to form what we observe as
broad personality dimensions. Jang et al. (2002) showed that two independent source of
genetic variance were necessary to explain the variation of each of the FFM personality
The evolutionary genetics of personality
37
domains. If these independent genetic sources reflect independent neurodevelopmental
mechanisms, environments may exist in which they no longer contribute to the same
behavioural dispositions (de Jong, 1990), and are no longer under parallel selection
pressures (e.g., see Figure 2b). The bottom line is that the genetic architecture of personality
might not reflect the phenotypic structure of established factor-analytic models, though it
would be surprising if it was completely different. At any rate, we believe that the reaction
norms of structurally independent mechanisms constitute a promising level of analysis for an
evolutionary personality psychology.
Operationalising Individual Reaction Norms
The natural approach to the study of reaction norms would be to observe the
behavioural reactions of different genotypes along a well-quantified environmental
continuum. However, the standard methods used by evolutionary geneticists to study nonhuman species (e.g. inbred strains) are of course not available to human psychologists.
Identical twins provide a surrogate (Crawford & Anderson, 1989), though a limited one, since
only two copies of each genotype exist and the environment cannot be varied experimentally.
One alternative is to relate single polymorphisms to behavioural variations that are
contingent on certain environmental variables (as done by Caspi et al., 2002, 2003, see also
Moffitt et al., 2006). While this approach will certainly become common in the near future as
a consequence of cheaper, faster, and more powerful genotyping methods (e.g. DNA
microarrays), such studies might still fail to capture the complex polygenic nature of
personality traits in the near future.
Another alternative is to assess individual differences directly at the level of
hypothetical underlying mechanisms. Here, an endophenotype approach appears highly
promising. Endophenotypes are phenotypic structures and processes (e.g. neurotransmitter
systems or hormone cascades) that can be quantified directly (e.g. by neuroimaging or blood
sampling) and that mediate between genes and more complex or abstract traits (Boomsma
et al., 1997; Cannon & Keller, 2005). In the watershed model (Figure 1), currently
The evolutionary genetics of personality
38
measurable endophenotypes tend to be located at a very upstream level. In the exemplary
case of neuroticism, amygdala reactivity (Hariri et al., 2002, 2005) provides an especially
good example of a mediating endophenotype, though there are likely several others. Sih et
al., (2004), for example, highlighted the role of hormonal mechanisms in animal personality.
Of course, all of these approaches are much harder work than using classical
personality questionnaires, so they will probably remain a minority interest within personality
psychology. But even questionnaires can be improved to reflect a view of traits as individual
reaction norms, by explicitly assessing behavioural reactions to specific fitness-relevant
situations, instead of aggregating across arbitrary modern environments (Mischel & Shoda,
1995; Denissen & Penke, 2006). For example, some people may be socially confident at
informal parties but not at public speaking, whereas for others, the opposite may apply. To
class them both as ‘extraverts’ may conflate disparate genotypes that lead to distinct
endophenotypes, behavioural strategies, reaction norms, and fitness payoffs. Indeed, the
quest to maximize internal consistencies within personality scales (e.g. by homogenizing the
environmental circumstances of behaviours) may lead personality psychologists to eliminate
some of the questionnaire items that are most informative about GxE interactions and
individual reaction norms.
An Evolutionary Genetic Model of Personality
The evolutionary genetics of personality can be summarized in the model depicted in
Figure 3.
Insert Figure 3 about here
For natural selection, the structure of individual differences is fairly straightforward
and simple: all living organisms vary on one major dimension - fitness – which is their
statistical propensity to pass their genes on to future generations to come. Miller’s (2000c) ffactor represents this dimension at the very top of any evolutionary hierarchy of heritable
differences – or at the very downstream end of the watershed model (which is why we put f
at the bottom in Figure 3). The upstream-downstream dimension is shown on the left. Since
The evolutionary genetics of personality
39
virtually all psychological differences studied so far show heritability, the central question for
evolutionary personality psychology is: how do psychological differences relate to the ffactor?
All heritable psychological differences begin with a set of genes that influence the
functioning of neurophysiological mechanisms (detectable as endophenotypes). A
simplification of the model is that environmental influences are omitted at the genetic and
endophenotype levels. This seems justifiable, since environmental effects are probably
smaller (due to developmental canalization) at the upstream levels than at the downstream
levels. One or several of the mechanisms on the endophenotype level result in the
behavioural tendencies that we observe as traits and abilities at the dispositional level. In
relevant situations, these dispositions influence behaviour, and from this point onward, they
affect the biological fate of the organism: behaviour influences the organism’s adaptive fit to
the current environment, and thus influences its overall reproductive success.
Genetic variation in personality differences might be maintained by selective
neutrality, mutation-selection balance, or balancing selection – each of which would leave
distinctive footprints in a trait’s genetic architecture. We have argued that selective neutrality
is implausible for most personality differences, given their pervasive effects on fitnessrelevant life outcomes. Mutation-selection balance requires that (1) a trait is influenced by
enough genes that new mutations disrupt its efficiency at a steady rate, and (2) selection
favours trait efficiency strongly enough to eliminate these mutations after some evolutionary
time. As a consequence, these traits will be influenced by many interdependent neurogenetic
mechanisms on the endophenotypic level, and will show substantial additive genetic variation
that affects trait efficiency and thereby influences fitness. Environmental influences on such
traits will be mediated mostly by their effects on the organism’s overall condition. In line with
Miller (2000c; Prokosch et al., 2005), we propose that general intelligence belongs to this
category of traits under mutation-selection balance. In this case, the upstream ability
mechanisms I and II in Figure 3 could be, for example, the efficiency of cerebral glucose
metabolism and the accuracy of prefrontal programmed cell death during adolescence, and
The evolutionary genetics of personality
40
the downstream ability mechanisms III and IV could be processing speed and working
memory capacity (see Jensen, 1998).
An evolutionary genetic conceptualisation of cognitive abilities would thus be:
individual differences in the functional integrity of broad systems of the adaptive cognitive
apparatus, caused by an individual’s load of rare, mildly harmful mutations. In short, cognitive
abilities are cognitive fitness components. For such traits, a low mutation load is always
beneficial, regardless of the environment.
By contrast, the phenotypic and genetic characteristics that are typically found in
studies of personality traits (like those in the FFM) suggest that balancing selection is
maintaining the genetic variance in most (if not all) personality traits. Balancing selection can
favour different traits in different social or non-social environments. In addition to this role as
a varying selection pressure on personality traits, the environment serves a second role
earlier on, when interacting with the neurophysiological architecture of the trait (i.e., its
personality mechanism or mechanisms) through a reaction norm to form a behavioural
tendency. This twofold role may make the environmental influences on personality traits
under balancing selection much more numerous, complex, and differentiated than those
affecting traits under mutation-selection balance (which may reflect general phenotypic
condition rather than specific environmental contingencies.) On the other hand, the upstream
genes and endophenotypes of personality traits under balancing selection will be fewer than
those of cognitive abilities under mutation-selection balance.
An evolutionary genetic conceptualisation of personality traits would thus be:
individual differences in genetic constraints on behavioural plasticity, which lead to
behavioural tendencies that follow individual reaction norms, and producing different fitness
consequences in different environments. In short, personality traits are individual reaction
norms with environment-contingent fitness consequences.
Where Do Common Psychopathologies Fit Into the Model?
The evolutionary genetics of personality
41
While this review focuses on personality differences in the normal range, we would
like to add some remarks on the place of polygenic psychopathologies in our model. In an
extensive discussion of the evolutionary genetics of common psychopathologies, Keller and
Miller (2006a, b) argued that mental disorders such as schizophrenia and bipolar disorder
are best conceptualized as traits under mutation-selection balance. Indeed, they cite
evidence that these disorders possess all the expected characteristics (Table 1). In our
model, these disorders are thus fitness components that mark the low end of the f-factor.
Some common psychopathologies, however, show clear relationships to personality traits in
the normal range, especially to high neuroticism and low agreeableness (Saulsman & Page,
2004). These disorders might be viewed as maladaptive extremes of normal personality traits
- rare genotypes that will sometimes occur in polygenic traits due to sexual recombination.
For example, extreme extraversion (e.g. impulsive, narcissistic, histrionic, and/or
promiscuous behaviour) and extreme introversion (e.g. schizoid, avoidant, hermit-like
withdrawal from all social contact) may both be too extreme to yield fitness benefits in any
plausible niche (MacDonald, 1995, 1998). But extreme values on normal traits alone are
usually insufficient for the occurrence of psychopathologies (Saulsman & Page, 2004), and
even high neuroticism and low agreeableness can be adaptive (though not necessarily
socially desirable) when the social environment is harsh, risky, and unforgiving, or when it is
exploitable and gullible, respectively (Nettle, 2006a; Denissen & Penke, 2006).
An alternative is that modern societies produce mismatches between heritable
temperaments and available niches. For example, Harpending and Cochran (2002) argue
that the very same 7R-DRD4 allele that predisposes children to attention deficit hyperactivity
disorder (ADHD) today may have been adaptive if these individuals lived in a violently
competitive, polygamous society. More generally, genetic variation maintained by
environmental heterogeneity implies that there are always some individuals for whom an
optimal niche does not currently exist. Similarly, negative frequency-dependent selection
implies that there are cases in which an individual’s usual niche is overcrowded and
competitive.
The evolutionary genetics of personality
42
In addition, the pathological nature of personality disorders might also result from a
high mutation load, but receive their characteristic symptoms from an interaction of this load
with certain personality traits. For example, very high openness to experience might
overwhelm individuals whose cognitive abilities are compromised by a high mutation load
and consequently lead to a diagnosed schizotypic personality disorder, while it might appear
attractive in less mutation-laden individuals, who are able to turn it into exceptional creative
outputs (see Nettle, 2006b; Nettle & Clegg, 2006; Keller & Miller, 2006b).
Practical Implications for Behaviour Genetics
An evolutionary genetic framework for personality psychology has some practical
implications for behaviour genetic studies:
(1) Demonstrating that a personality trait is heritable had become scientifically
unsurprising by the early 1990s (Turkheimer & Gottesman, 1991), and is not very
informative about a trait’s nature or etiology, since it confounds information about
a trait’s evolutionary history, structure, and GxE interactions (Stirling et al., 2002).
This is especially true for the broad-sense heritabilities that are estimated in the
classical twin design, since they do not distinguish between VA and VNA (Keller &
Coventry, 2005), which is very important in evolutionary genetics (Merliä &
Sheldon, 1999). We therefore concur with Keller and Coventry (2005) that more
studies using the extended twin-family design (Neale & Maes, 2004), or other
designs that unconfound VA and VNA, are highly desirable, especially when testing
evolutionary genetic hypotheses (cp. Table 1).
(2) Because of the great datasets and twin registries already available, classical twin
studies will probably remain the most common type of behaviour genetic
publications. However, such studies would be more informative (or less
misleading) about the evolutionary genetics of traits if their underlying statistical
assumptions were made more explicit. Many personality psychologists seem not
to appreciate that classical twin studies can yield a wide range of mathematically
The evolutionary genetics of personality
43
equivalent parameter estimates (e.g. for additive genetic vs. dominance vs.
epistatic effects) that have very different implications for the evolutionary histories
of the traits under investigation (Keller & Coventry, 2005; Coventry & Keller,
2005). We therefore suggest that future publications of classical twin study results
make use of the technique developed by Keller and Coventry (2005) and fully
disclose the confidence intervals and parameter spaces for their results.
(3) The equation of personality differences with individual reaction norms highlights
the fact that GxE interactions are ubiquitous in nature. Similarly, balancing
selection on personality traits due to spatiotemporal heterogeneity of selection
pressures suggests that GxE correlations are fairly common. Unfortunately, the
usual approach in quantitative behaviour genetic studies is additive variance
decomposition, which hides both GxE interactions and GxE correlations in
apparent main effects (Purcell, 2002). However, the necessary statistical
modelling techniques exist to identify such interaction effects (Neale & Maes,
2004; Purcell, 2002), and evolutionary genetics suggests that they should be used
more frequently.
(4) For the same reason, the use of personality trait measures (especially self-report
questionnaires) that aggregate across situations might have reached its limits in
clarifying the genetic architecture of personality (Ebstein, 2006). Both
endophenotype approaches and phenotypic measures that aim to keep person
and situation separated (Dennisen & Penke, 2006; Mischel & Shoda, 1995)
provide better alternatives.
(5) Calculating the coefficient of additive genetic variance (CVA) of a trait, which is
very informative about its evolutionary history (Houle, 1992; Stirling et al., 2002),
requires a ratio-scale measure (i.e., a measure with a meaningful zero point).
Personality questionnaires with rating scales fail to reach this standard. It would
be very helpful if valid, ratio-scaled personality measures (e.g. based on
quantitative endophenotypes or behaviours measured with regard to their energy
The evolutionary genetics of personality
44
output, temporal duration, or act frequency – see Buss & Craik, 1993) could be
developed and used in quantitative behaviour genetic studies.
(6) We predict that ‘gene hunting’ studies will continue to be more successful in
revealing the molecular genetic architecture of temperamental personality traits
than of general cognitive abilities or polygenic mental disorders (Ebstein, 2006;
Plomin et al., 2006; Keller & Miller, 2006a). Evolutionary genetic theory gives a
straightforward reason why: while personality traits will be influenced by a limited
set of high-prevalence alleles (plus maybe several rare ones, see Kopp &
Hermisson, 2006), general intelligence and psychopathologies like schizophrenia
will be influenced by rare, recessive, mildly harmful mutations that vary between
samples, since they are equally likely to occur at thousands of different, otherwise
monomorphic loci, and are removed fairly quickly by selection once they arise.
(Note that this goes beyond Kovas and Plomin’s (2006) concept of ‘generalist
genes’, which proposes that the same large set of weak-effect polymorphisms
underlies cognitive functioning in every individual.) While we do not argue that
molecular behavioural geneticists should refrain from studying g, common
psychopathologies, and other fitness components, we suggest that they take
evolutionary genetic predictions of the likely genetic architecture into account
when planning studies and interpreting results. A simple first step would be to call
the underlying polymorphisms what the empirical evidence suggests they are –
rare mutations.
(7) More generally, evolutionary genetics provides a rich theoretical source of
hypotheses that should inspire and guide future behaviour genetic studies. For
example, factor V (openness to experiences/intellect) is the only domain of the
FFM that shows reliable correlations with general intelligence (e.g. DeYoung et
al., 2005). From an evolutionary genetics viewpoint, this puts factor V in an
ambiguous position: does it reflect an ESS under balancing selection (see
Denissen & Penke, 2006; Nettle, 2006a), or an important component of the f-
The evolutionary genetics of personality
45
factor, which should be under mutation-selection balance? If factor V is under
balancing selection, its molecular genetic basis should be much easier to identify
– especially if behaviour genetics researchers statistically control for general
intelligence when investigating polymorphisms that may influence factor V. Other
exemplary evolutionary genetic hypotheses can be found in Miller (2000c) and
Keller (in press).
Conclusion
Evolutionary psychology has made so much progress in the last 15 years by relying
on an evolutionary adaptationist metatheory that guides the identification of ancestral
adaptive problems, the likely psychological adaptations that they favoured, and the likely
design features of those adaptations that can be investigated empirically (Andrews et al.,
2002; Buss, 1995). We have argued that evolutionary genetics can provide a similarly
powerful approach to the study of heritable individual differences in personality.
Evolutionary genetics is itself a fast evolving field. While we tried to give an up-to-date
overview of evolutionary genetic principles that seemed most relevant for personality
psychology, some of those principles will probably be refined, extended, or challenged in the
near future. They should thus be viewed as the provisional, current state of the art, not as
biological commandments carved in stone. Still, they may help personality psychology
enormously by clarifying what is evolutionarily possible and plausible, and what is not. This
way, evolutionary genetics can provide personality psychology with new hypotheses,
guidance on how to interpret results, and constraints on theory formulation. Ultimately, our
grandest hope for evolutionary personality psychology is that, given the enormously rich
phenotypic and behaviour genetic datasets on human personality, it might identify new
evolutionary genetic principles that also apply to other kinds of traits and other species.
We reviewed the current answers that evolutionary genetics can give to a question
that has rarely been asked in psychology: how is the genetic variation that obviously
underlies most human differences, including personality differences, maintained in the
The evolutionary genetics of personality
46
population? It turned out that only two answers are sufficiently plausible for personality
differences: either (1) the trait is dependent on so many genes that a balance between rare,
mildly harmful mutations and counteracting selection occurs, or (2) variation in the structure
of the physical or social environment leads to spatiotemporally fluctuating selection for
different alleles. Both evolutionary genetic mechanisms will lead their affected traits to have
certain distinctive characteristics and underlying genetic architectures. We concluded that the
first process (mutation-selection balance) probably maintains genetic variance in cognitive
abilities, while the second process (balancing selection by environmental heterogeneity)
probably maintains genetic variance in most personality traits. Thus, cognitive abilities are
best conceptualised as cognitive fitness components, while personality traits reflect individual
reaction norms with environment-contingent fitness consequences.
Important tasks for future studies include delineating the hierarchical structure of
fitness components (with the f–factor on the top) and identifying the exact fitness-related
costs and benefits associated with each personality trait, as well as the environmental niches
that structure those costs and benefits. Social niches with different degrees and forms of
competition are especially good candidates for the latter. A promising road for processoriented personality psychologists is studying the psychological mechanisms that lead to
active niche selection, including adaptive self-assessments (Tooby & Cosmides, 1990;
Penke & Denissen, 2007; Penke et al., in press) and experience-producing drives (Bouchard
et al., 1996).
Finally, we wish to reemphasise that most heritable individual differences are not
adaptations in their own right. They reflect dimensions in the functional design of a species
that tolerates some degree of genetic variation. Mutations at too many non-neutral loci will
lead to a breakdown of adaptive design. Likewise, traits under balancing selection will
tolerate polymorphisms only at a few specific loci, while all others loci (which affect the
universal adaptative design of the trait) will be protected from large genetic variation by
stabilizing selection. Adaptive individual differences exist, but only as conditional strategies
that are implemented in universal (i.e. zero-heritability) adaptations and evoked by specific
The evolutionary genetics of personality
47
environmental cues (Tooby & Cosmides, 1990; Buss, 1991; Buss & Greiling, 1999). An
evolutionary personality psychology based on evolutionary genetics does not contradict this
view. Instead, it complements evolutionary psychology by explaining what happens when
genetic variation is introduced into systems of interacting adaptations (Gangestad & Yeo,
1997; Miller, 2000a). Since genetic variation is ubiquitous in personality psychology,
evolutionary genetics is essential for an evolutionary personality psychology.
The evolutionary genetics of personality
48
References
Alexander, R. D. (1989). Evolution of the human psyche. In P. Mellars, & C. Stringer (Eds.),
The human revolution: behavioural and biological perspectives on the origins of modern
humans (pp. 455–513). Princeton: University Press.
Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt.
Andrews, P. W., Gangestad, S. W., & Matthews, D. (2002). Adaptationism - How to carry out
an exaptationist program. Behavioral and Brain Sciences, 25, 489-504.
Ashton, M. C., & Lee, K. (2001). A theoretical basis for the major dimensions of personality.
European Journal of Personality, 15, 327-353.
Barrett, H. C. (2006). Modularity and design reincarnation. In P. Carruthers, S. Laurence & S.
Stich (Eds.), The innate mind: Culture and cognition. Oxford: University Press.
Bates, T. C. (2007). Fluctuating asymmetry and intelligence. Intelligence, 35, 41-46.
Boomsma, D. I., Anokhin, A., & De Geus, E. J. C. (1997). Genetics of electrophysiology:
Linking genes, brains, and behavior. Current Directions in Psychological Science, 6, 106110.
Borkenau, P., Riemann, R., Spinath, F. M., & Angleitner, A. (2006). Genetic and
environmental influences on person x situation profiles. Journal of Personality, 74, 14511479.
Bouchard, T. J., jr., Lykken, D. T., Tellegen, A., & McGue, M. (1996). Genes, drives,
environment, and experience: EPD theory revised. In C. Benbow, & D. J. Lubinski (Eds.),
Intellectual talent: Psychometric and social issues (pp. 5-43). Baltimore: John Hopkins
University Press.
Brcic-Kostic, K. (2005). Neutral mutation as the source of genetic variation in life history
traits. Genetical Research, 86, 53-63.
Brunswik, E. (1956). Perception and the representative design of experiments. Berkely:
University of California Press.
Bürger, R. (2000). The mathematical theory of selection, recombination, and mutation. West
Sussex: Wiley.
Bürger, R (2005) A multilocus analysis of intraspecific competition and stabilizing selection
on a quantitative trait. Journal of Mathematical Biology, 50, 355-396
Burt, A. (1995). The evolution of fitness. Evolution, 49, 1-8.
Buss, D. M. (1990). Toward a biologically informed psychology of personality. Journal of
Personality, 58, 1-16.
Buss, D. M. (1991). Evolutionary personality psychology. Annual Review of Psychology, 42,
459-491.
Buss, D. M. (1995). Evolutionary psychology: A new paradigm for psychological science.
Psychological Inquiry, 6, 1-30.
Buss, D. M., & Craik, K. H. (1983). The act frequency approach to personality. Psychological
Review, 90, 105-126.
Buss, D. M., & Greiling, H. (1999). Adaptive individual differences. Journal of Personality, 67,
209-243.
Cannon, T. D. & Keller, M. C. (2005) Endophenotypes in genetic analyses of mental
disorders. Annual Review of Clinical Psychology, 2, 267–90.
Cargill, M., Altshuler, D., Ireland, J., Sklar, P., Ardlie, K., Patil, N., et al. (1999).
Characterization of single-nucleotide polymorphisms in coding regions of human genes.
Nature Genetics, 22, 231-238.
Caspi, A., McClay, J., Moffitt, T. E., Mill, J., Martin, J., Craig, I. W., Taylor, A., & Poulton, R.
(2002). Role of genotype in the cycle of violence in maltreated children. Science, 297,
851-854.
Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington, H. L., McClay, J.,
Mill, J., Martin, J., Braithwaite, A., & Poulton, R. (2003). Influence of life stress on
depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301, 386-389.
Cervone, D. (2000). Evolutionary psychology and explanation in personality psychology: How
do we know which module to invoke? American Behavioral Scientist, 43, 1001-1014.
The evolutionary genetics of personality
49
Chang, F. M., Kidd, J. R. , Livak, K. J., Pakstis, A. J., & Kidd, K. K. (1996). The worldwide
distribution of allele frequencies at the human dopamine D4 receptor locus. Human
Genetics, 98, 91-101.
Chen, C., Burton, M., Greenberger, E., & Dmitrieva, J. (1999). Population migration and the
variation of dopamine D4 receptor (DRD4) allele frequencies around the globe. Evolution
and Human Behavior, 20, 309-324.
Chipuer, H. M., Rovine, M. J., & Plomin, R. (1990). LISREL modeling: Genetic and
environmental influences on IQ revisited. Intelligence, 14, 11–29.
Ciani, A. S. C., Capiluppi, C., Veronese, A., & Sartori, G. (in press). The adaptive value of
personality differences revealed by small island population dynamics. European Journal
of Personality.
Coventry, W. L., & Keller, M. C. (2005). Estimating the extent of parameter bias in the
classical twin design: A comparison of parameter estimates from extended twin-family
and classical twin designs. Twin Research and Human Genetics, 8, 214-223.
Cosmides, L., & Tooby, J. (2002). Unraveling the enigma of human intelligence: Evolutionary
psychology and the multimodular mind. In R. J. Sternberg & J. C. Kauman (Eds.): The
evolution of intelligence (pp. 145-198). Mahwah: Erlbaum.
Crawford, C. B., & Anderson, J. L. (1989). Sociobiology: An environmentalist discipline?
American Psychologist, 44, 1449-1459.
Crnokrak, P., & Roff, D.A. (1995). Dominance variance: Associations with selection and
fitness. Heredity, 75, 530–540.
Cronbach, L. J. (1949). Essentials of psychological testing. New York: Harper.
Curtisinger, J. W., Service, P. M., & Prout, T. (1994). Antagonistic pleiotropy, reversal of
dominance, and genetic polymorphism. The American Naturalist, 144, 210-228.
Darwin, C. (1859). On the Origin of Species. London: John Murray.
Deary, I. J., & Der, G. (2005). Reaction time explains IQ’s association with death.
Psychological Science, 16, 64-69.
Deary, I. J., Whiteman, M. C., Starr, J. M., Whalley, L. J., & Fox, H. C. (2004). The impact of
childhood intelligence on later life: Following up the Scottish mental surveys of 1932 and
1947. Journal of Personality and Social Psychology, 86, 130-147.
de Jong, G. (1990). Quantitative genetics of reaction norms. Journal of Evolutionary Biology,
3, 447-468.
Denissen, J. J. A., & Penke, L. (2006). Individual reaction norms underlying the Five Factor
Model of personality: First steps towards a theory-based conceptual framework.
Manuscript submitted for publication.
DeRose, M. A., & Roff D. A. 1999. A comparison of inbreeding depression in life history and
morphological traits in animals. Evolution, 53, 1288–1292.
DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2005). Sources of openness/intellect:
Cognitive and neuropsychological correlates of the fifth factor of personality. Journal of
Personality, 73, 825-858.
Ding, Y.-C., Chi, H.-C., Grady, D. L., Morishima, A., Kidd, J. R., Kidd, K. K., et al. (2002).
Evidence of positive selection acting at the human dopamine receptor D4 gene locus.
Proceedings of the National Academy of Sciences of the United States of America, 99,
309-314.
Dingemanse, N. J., Both, C., Drent, P. J., & Tinbergen, J. M. (2004). Fitness consequences
of avian personalities in a fluctuating environment. Proceedings of the Royal Society of
London, Series B, 271, 847-852.
Dingemanse, N. J., & Reale, D. (2005). Natural selection and animal personality. Behaviour,
142, 1159-1184.
Eaves, L., Heath, A., Martin, N., Maes, H., Neale, M., Kendler, K., et.al. (1999). Comparing
the biological and cultural inheritance of personality and social attitudes in the Virginia 30
000 study of twins and their relatives. Twin Research, 2, 62–80.
Eaves, L. J., Heath, A. C., Neale, M. C., Hewitt, J. K., & Martin, N. G. (1998). Sex differences
and non-additivity in the effects of genes on personality. Twin Research, 1, 131–137.
Eaves, L. J., Martin, N. G., Heath, A. C., Hewitt, J. K., & Neale, M. C. (1990). Personality and
reproductive fitness. Behavior Genetics, 20, 563-568.
The evolutionary genetics of personality
50
Ebstein, R. (2006). The molecular genetic architecture of human personality: Beyond selfreport questionnaires. Molecular Psychiatry, 11, 427-445.
Ellis, B. J., Simpson, J. A., & Campbell, L. (2002). Trait-specific dependence in romantic
relationships. Journal of Personality, 70, 611-659.
Endler, J. A. (1986). Natural selection in the wild. Princeton: University Press.
Ewens, W. J. (1989). An interpretation and proof of the fundamental theorem of natural
selection. Theoretical Population Biology, 36, 167–180.
Eyre-Walker, A., & Keightley, P. D. (1999). High genomic deleterious mutation rates in
hominids. Nature, 397, 344-347.
Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual differences: A natural
science approach. Plenum: New York.
Falconer, D. S. (1981). Introduction to quantitative genetics (2nd ed.). New York: Longman.
Fay, J. C., Wyckoff, G. J., & Wu, C. (2001). Positive and negative selection on the human
genome. Genetics, 158, 1227-1234.
Figueredo, A. J., Hammond, K. R., & McKiernan, E. C. (2006). A Brunswikian evolutionary
developmental theory of preparedness and plasticity. Intelligence, 34, 211-227.
Figueredo, A. J., Sefcek, J. A., & Jones, D. N. (2006). The ideal romantic partner personality.
Personality and Individual Differences, 41, 431-441.
Figueredo, A. J., Sefcek, J. A., Vasquez, G., Brumbach, B. H., King, J. E., & Jacobs, W. J.
(2005a). Evolutionary personality psychology. In D. M. Buss (Ed.), The handbook of
evolutionary psychology (pp. 851-877). Hoboken, NJ, US: Wiley.
Figueredo, A. J., Vasquez, G., Brumbach, B. H., Sefcek, J. A., Kirsner, B. R., & Jacobs, W.
(2005b). The K-factor: Individual differences in life history strategy. Personality and
Individual Differences, 39, 1349-1360.
Fisher, R. A. (1930). The genetical theory of natural selection. Oxford: Clarendon Press.
Friedman, H. S., Tucker, J. S., Schwartz, J. E., Martin, L. R., Tomlinsonkeasey, C., Wingard,
D. L., & Criqui, M. H. (1995). Childhood conscientiousness and longevity: Health
behaviors and cause of death. Journal of Personality and Social Psychology, 68, 696703.
Gangestad, S. W., & Simpson, J. A. (1990). Toward an evolutionary history of female
sociosexual variation. Journal of Personality, 58, 69-96.
Gangestad, S. W., & Simpson, J. A. (2000). The evolution of human mating: Trade-offs and
strategic pluralism. Behavioral and Brain Sciences, 23, 573-644.
Gangestad, S. W., & Yeo, R. A. (1997). Behavioral genetic variation, adaptation and
maladaptation: an evolutionary perspective. Trends in Cognitive Sciences, 1, 103-108.
Garcia-Dorado, A., Caballero, A., & Crow, J. F. (2003). On the persistence and
pervasiveness of a new mutation. Evolution, 57, 2644-2646.
Garrigan, D., & Hedrick, P. W. (2003). Detecting adaptive molecular polymorphism: Lessons
from the MHC. Evolution, 57, 1707-1722.
Gigerenzer, G., Todd, P. M., & the ABC Research Group (Eds.). (1999). Simple heuristics
that make us smart. New York: Oxford University Press.
Goldberg, L. R. (1981). Language and individual differences: The search for universals in
personality lexicons. In L. Wheeler (Ed.), Review of personality and social psychology,
Vol. 2 (pp. 141-165). Beverly Hills: Sage.
Gottfredson, L. S. (2004). Intelligence: Is it the epidemiologists’ elusive “fundamental cause”
of social class inequalities in health? Journal of Personality and Social Psychology, 86,
174-199.
Gottfredson, L. S. (in press). Innovation, fatal accidents, and the evolution of general
intelligence. In M. J. Roberts (Ed.), Integrating the mind. Hove: Psychology Press.
Hariri, A. R., Drabant, E. M., Munoz, K. E., Kolachana, B. S., Mattay, V. S., Egan, M. F., et al.
(2005). A susceptibility gene for affective disorders and the response of the human
amygdala. Archives of General Psychiatry, 62, 146–152.
Hariri, A. R., Mattay, V. S., Tessitore, A., Kolachana, B., Fera, F., Goldman, D., et al. (2002).
Serotonin transporter genetic variation and the response of the human amygdala.
Science, 297, 400–403.
The evolutionary genetics of personality
51
Harpending, H., & Cochran, G. (2002). In our genes. Proceedings of the National Academy
of Sciences USA, 99, 10-12.
Hedrick, P. W. (1999). Antagonistic pleiotropy and genetic polymorphism: A perspective.
Heredity, 82, 126-133.
Hogan, R. (1996). A socioanalytic perspective on the five-factor model. In J. S. Wiggins
(Ed.), The five factor model of personality: Theoretical perspectives (pp. 163-179). New
York: Guilford Press.
Houle, D. (1992). Comparing evolvability and variability of quantitative traits. Genetics, 130,
195-204.
Houle, D. (1998). How should we explain variation in the genetic variance of traits? Genetica,
102/103, 241-253.
Houle, D., Hughes, K. A., Hoffmaster, D. K., Ihara, J., Assimacopolous, S., Canada, D., &
Charlesworth, B. (1994). The effects of spontaneous mutation on quantitative traits I:
Variance and covariance of life history traits. Genetics, 138, 773-785.
Human Genome Project, 2001. Insights learned from the sequence. Retrieved September
2nd, 2006,
http://www.ornl.gov/sci/techresources/Human_Genome/project/journals/insights.html.
Hughes, K. A., & Burleson, M. H. (2000). Evolutionary causes of the variation in fertility and
other fitness traits. In J. L. Rodgers, D. C. Rowe & W. Miller (Eds.), Genetic influences on
human sexuality and fertility. Dordrecht: Kluwer.
Hurles, M. (2004). Gene duplication: The genomic trade in spare parts. PLoS Biology, 2,
900-904.
International HapMap Consortium (2005). A haplotype map of the human genome. Nature,
437, 1299–1320.
James, W. (1890). The principles of psychology. New York: Holt.
Jang, K. L., Livesley, W. J., Ando, J., Yamagata, S., Suzuki, A., Angleitner, A., et al. (in
press). Behavioral genetics of the higher-order factors of the Big Five. Personality and
Individual Differences.
Jang, K. L., Livesley, W., Angleitner, A., Riemann, R., & Vernon, P. A. (2002). Genetic and
environmental influences on the covariance of facets defining the domains of the fivefactor model of personality. Personality and Individual Differences, 33, 83-101.
Jang, K. L., Livesley, W., Riemann, R., Vernon, P. A., Hu, S., Angleitner, A., et al. (2001).
Covariance structure of neuroticism and agreeableness: A twin and molecular genetic
analysis of the role of the serotonin transporter gene. Journal of Personality and Social
Psychology, 81, 295-304.
Jang, K. L., McCrae, R. R., Angleitner, A., Riemann, R., & Livesley, W. (1998). Heritability of
facet-level traits in a cross-cultural twin sample: Support for a hierarchical model of
personality. Journal of Personality and Social Psychology, 74, 1556-1565.
Jensen, A. R. (1998). The g-factor. The science of mental ability. Westport: Praeger.
John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement and
theoretical perspectives. In L. A. Pervin & O. P. John (Eds.), Handbook of personality:
Theory and Research (pp. 102-138). New York: Guilford Press.
Kanazawa, S. (2004). General intelligence as a domain-specific adaptation. Psychological
Review, 111, 512-523.
Kassen, R. (2002). The experimental evolution of specialists, generalists, and the
maintenance of diversity. Journal of Evolutionary Biology, 15, 173-190.
Keightley, P. D., & Gaffney, D. J. (2003). Functional constraints and frequency of deleterious
mutations in noncoding DNA of rodents. Proceedings of the National Academy of
Sciences of the United States of America, 100, 13402-13406.
Keller, M. C. (in press). The role of mutations in human mating. In G. Geher & G. Miller
(Eds.), Mating intelligence: New insights into intimate relationships, human sexuality, and
the mind’s reproductive system. Mahwah: Lawrence Erlbaum.
Keller, M. C., & Coventry, W. L. (2005). Quantifying and addressing parameter indeterminacy
in the classical twin design. Twin Research and Human Genetics, 8, 201-213.
The evolutionary genetics of personality
52
Keller, M. C., Coventry, W. L., Heath, A. C., & Martin, N. G. (2005). Widespread evidence for
non-additive genetic variation in Cloninger’s and Eysenck’s personality dimensions using
a twin plus sibling design. Behavior Genetics, 35, 707-721.
Keller, M. C., & Miller, G. F. (2006a). Resolving the paradox of common, harmful, heritable
mental disorders: Which evolutionary genetic models work best? Behavioral and Brain
Sciences, 29, 385-404.
Keller, M. C., & Miller, G. F. (2006b). An evolutionary framework for mental disorders:
Integrating adaptationist and evolutionary genetic models. Behavioral and Brain
Sciences, 29, 429-441.
Kidd, K. K. (2006). ALFRED – the ALlele FREquency Database. Retrieved September 2nd,
2006, http://alfred.med.yale.edu.
Kimura, M. (1983). The neutral theory of molecular evolution. Cambridge, U.K.: Cambridge
University Press.
Kokko, H., Brooks, R., Jennions, M. D., & Morley, J. (2003). The evolution of mate choice
and mating biases. Proceedings of the Royal Society of London, Series B, 270, 653-664.
Kopp, M., & Hermisson, J. (2006). Evolution of genetic architecture under frequencydependent disruptive selection. Evolution, 60, 1537-1550.
Kovas, Y., & Plomin, R. (2006). Generalist genes: implications for the cognitive sciences.
Trends in Cognitive Sciences, 10, 198-203.
Krebs, J. R., & Davies, N. B. (1997). Behavioural ecology (4th ed.). Oxford: Blackwell
Science.
Luxen, M. F., & Buunk, B. P. (2006). Human intelligence, fluctuating asymmetry and the
peacock's tail: General intelligence (g) as an honest signal of fitness. Personality and
Individual Differences, 41, 897-902.
Lykken, D. T., & Tellegen, A. (1993). Is human mating adventitious or the result of lawful
choice? A twin study of mate selection. Journal of Personality and Social Psychology, 65,
56-68.
Lynch, M., & Hill, W. G. (1986). Phenotypic evolution by neutral mutation. Evolution, 40, 915935.
Lynch, M., & Walsh, B. (1998). Genetics and analysis of quantitative traits. Sunderland:
Sinauer.
MacDonald, K. (1995). Evolution, the five-factor model, and levels of personality. Journal of
Personality, 63, 525-567.
MacDonald, K. (1998). Evolution, culture, and the five-factor model. Journal of Cross Cultural
Psychology, 29, 119-149.
Marcus, G. (2004). The birth of the mind: How a tiny number of genes creates the
complexities of human thought. New York: Basic Books.
Maynard Smith, J. (1982). Evolution and the theory of games. Cambridge: University Press.
Maynard Smith, J. (1998). Evolutionary genetics. Oxford: University Press.
Mayr, E. (1993). What was the evolutionary synthesis? Trends in Ecology and Evolution, 8,
31-34.
McAdams, D. P., & Pals, J. L. (2006). A new Big Five: Fundamental principles for an
integrative science of personality. American Psychologist, 61, 204-217.
McCrae, R. R., & Costa, P. T. (1996). Toward a new generation of personality theories:
Theoretical contexts for the five-factor model. In J. S. Wiggins (Ed.), The five factor model
of personality: Theoretical perspectives (pp. 51-87). New York: Guilford Press.
McCrae, R. R., Jang, K. L., Livesley, W., Riemann, R., & Angleitner, A. (2001). Sources of
structure: Genetic, environmental, and artifactual influences on the covariation of
personality traits. Journal of Personality, 69, 511-535.
McDougall, W. (1908). Introduction to Social Psychology. London: Methuen.
Mealey, L. (1995). The sociobiology of sociopathy: An integrated evolutionary model.
Behavioral and Brain Sciences, 18, 523-599.
Merilä, J., & Sheldon, B. C. (1999). Genetic architecture of fitness and nonfitness traits:
empirical patterns and development of ideas. Heredity, 83, 103-109.
The evolutionary genetics of personality
53
Miller, G. F. (2000a). Mental traits as fitness indicators: Expanding evolutionary psychology’s
adaptationism. In D. LeCroy & P. Moller (Eds.), Evolutionary perspectives on human
reproductive behavior, pp. 62-74.
Miller, G. F. (2000b). The mating mind: How sexual choice shaped the evolution of human
nature. New York: Doubleday.
Miller, G. F. (2000c). Sexual selection for indicators of intelligence. In G. Bock, J. Goode, &
K. Webb (Eds.), The nature of intelligence (pp. 260-275). New York: John Wiley.
Miller, G. F. (in press). Reconciling ecological psychology and evolutionary psychology: How
to perceive fitness affordances. Acta Psychologica Sinica.
Miller, G. F., & Penke, L. (in press). The evolution of human intelligence and the coefficient of
additive genetic variance in human brain size. Intelligence.
Mischel, W. (2004). Toward an integrative science of the person. Annual Review of
Psychology, 55, 1-22.
Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality:
Reconceptualizing situations, dispositions, dynamics, and invariance in personality
structure. Psychological Review, 102, 246-268.
Moffitt, T. E., Caspi, A., & Rutter, M. (2006). Measured Gene-Environment Interactions in
Psychopathology: Concepts, Research Strategies, and Implications for Research,
Intervention, and Public Understanding of Genetics. Perspectives on Psychological
Science, 1, 5-27.
Møller, A. P. (1997). Developmental stability and fitness: A review. American Naturalist, 149,
916-932.
Neale, M. C., & Cardon, L. R. (2004). Methodology for genetic studies for twins and families.
Dordrecht: Kluwer.
Neeleman, J., Sytema, S., & Wadsworth, M. (2002). Propensity to psychiatric and somatic illhealth: evidence from a birth cohort. Psychological Medicine, 32, 793-803.
Nettle, D. (2005). An evolutionary approach to the extraversion continuum. Evolution and
Human Behavior, 26, 363-373.
Nettle, D. (2006a). The evolution of personality variation in humans and other animals.
American Psychologist, 61, 622-631.
Nettle, D. (2006b). Reconciling the mutation-selection balance model with the schizotypycreativity connection. Behavioural and Brain Sciences, 29, 418.
Nettle, D. and Clegg, H. (2006). Schizotypy, creativity and mating success in humans.
Proceedings of the Royal Society, Series B, 273, 611-5.
Ozer, D. J., & Benet-Martinez, V. (2006). Personality and the prediction of consequential
outcomes. Annual Review of Psychology, 57, 401-421.
Penke, L., & Denissen, J. J. A. (2007). Adaptive sociometer contingencies and the allocation
of reproductive effort in men. Manuscript submitted for publication.
Penke, L., Todd, P. M., Lenton, A., & Fasolo, B. (in press). How self-assessments can guide
human mating decisions. In G. Geher & G. F. Miller (Eds.), Mating Intelligence: New
insights into intimate relationships, human sexuality, and the mind’s reproductive system.
Mahwah: Lawrence Erlbaum.
Pigliucci, M. (2005). Evolution of phenotypic plasticity: where are we going now? Trends in
Ecology and Evolution, 20, 481-486.
Plomin, R., DeFries, J. C., McClearn, G. E., & McGuffin, P. (2001). Behavioral genetics (4th
ed.). New York: Worth Publishers.
Plomin, R., Kennedy, J. K. J., Craig, I. W. (2006). The quest for quantitative trait loci
associated with intelligence. Intelligence, 34, 513-526.
Plomin, R., Owen, M. J., McGuffin, P. (1994). The genetic basis of complex human
behaviors. Science, 264, 1733-1739.
Plomin, R., & Spinath, F. M. (2004). Intelligence: Genetics, genes, and genomics. Journal of
Personality and Social Psychology, 86, 112–129.
Polak, M. (Ed.) (2003). Developmental instability: Causes and consequences. Oxford:
University Press.
Pomiankowski, A., & Møller, A. P. (1995). A resolution to the lek paradoxon. Proceedings of
the Royal Society London, Series B, 260, 21-29.
The evolutionary genetics of personality
54
Price, G. R. (1972). Fisher’s ‘fundamental theorem’ made clear. Annals of Human Genetics,
36, 129–140.
Price, T., & Schluter, D. (1991). On the low heritability of life-history traits. Evolution, 45, 853861.
Pritchard, J. K. (2001). Are rare variants responsible for susceptibility to complex diseases?
American Journal of Human Genetics, 69, 124-137.
Prokosch, M. D., Yeo, R. A., & Miller, G. F. (2005). Intelligence tests with higher g-loadings
show higher correlations with body symmetry: Evidence for a general fitness factor
mediated by developmental stability. Intelligence, 33, 203-213.
Purcell, S. (2002). Variance components models for gene environment interaction in twin
analysis. Twin Research, 5, 554–571.
Redon, R., Ishikawa, S., Fitch, K. R., Feuk, L., Perry, G. H., Andrews, T. D., et al. (2006).
Global variation in copy number in the human genome. Nature, 444, 444-454.
Rice, W. R., & Chippindale, A. K. (2001). Intrasexual ontogenetic conflict. Journal of
Evolutionary Biology, 14, 685-693.
Ridley, M. (2000). Mendel's demon: Gene justice and the complexity of life. London: Orion.
Roff, D. A. (1997). Evolutionary quantitative genetics. New York: Chapman & Hall.
Rowe, L., & Houle, D. (1996). The lek paradoxon and the capture of genetic variance by
condition dependent traits. Proceedings of the Royal Society London, Series B, 263,
1415-1421.
Rudan, I., Smolej-Narancic, N., Campbell, H., Carothers, A., Wright, A., Janicijevic, B., et al.
(2003). Inbreeding and the genetic complexity of human hypertension. Genetics, 163,
1011-1021.
Rushton, J. P. (1985). Differential K theory: The sociobiology of individual and group
differences. Personality and Individual Differences, 6, 441–452.
Saulsman, L. M., & Page, A. C. (2004). The five-factor model and personality disorder
empirical literature: A meta-analytic review. Clinical Psychology Review, 23, 1055-1085.
Schmitt, D. P. (2005). Sociosexuality from Argentina to Zimbabwe: A 48-nation study of sex,
culture, and strategies of human mating. Behavioral and Brain Sciences, 28, 247-275.
Schneider, K. A. (2006). A multilocus-multiallele analysis of frequency-dependent selection
induced by intraspecific competition. Journal of Mathematical Biology, 52, 483-523.
Shapiro, J. A., & von Sternberg, R. (2005). Why repetitive DNA is essential to genome
function. Biological Reviews, 80, 227-250.
Sih, A., Bell, A. M., Johnson, J. C., & Ziemba, R. E. (2004). Behavioral syndromes: An
integrative overview. Quarterly Review of Biology, 79, 241-277.
Smith, B. R., & Blumstein, D. T. (2007). Personality affects reproductive success, not
survival: A meta-analytic review. Manuscript submitted for publication.
Stirling, D. G., Reale, D., & Roff, D. A. (2002). Selection, structure and the heritability of
behaviour. Journal of Evolutionary Biology, 15, 277-289.
Strobel, A., Lesch, K., Jatzke, S., Paetzold, F., & Brocke, B. (2003). Further evidence for a
modulation of novelty seeking by DRD4 exon III, 5-HTTLPR, and COMT val/met variants.
Molecular Psychiatry, 8, 371-372.
Sunyaev, S., Ramensky, R., Koch, I., Lathe, W., Kondrashov, A. S., & Bork, P. (2001).
Prediction of deleterious human alleles. Human Molecular Genetics, 10, 591-597.
Thorndike, E. L. (1909). Darwin's contribution to psychology. University of California
Chronicle, 12, 65-80.
Tomkins, J. L., Radwan, J., Kotiaho, J. S., & Tregenza, T. (2004). Genic capture and
resolving the lek paradox. Trends in Ecology and Evolution, 19, 323-328.
Tooby, J., & Cosmides, L. (1990). On the universality of human nature and the uniqueness of
the individual: The role of genetics and adaptation. Journal of Personality, 58, 17-67.
Tooby, J., & Cosmides, L. (2005). Conceptual foundations of evolutionary psychology. In D.
M. Buss (Ed.), The handbook of evolutionary psychology (pp. 5-67). Hoboken: Wiley.
Tooby, J., Cosmides, L., & Barrett, H. C. (2005). Resolving the debate on innate ideas:
Learnability constraints and the evolved interpenetration of motivational and conceptual
functions. In P. Carruthers, S. Laurence & S. Stich (Eds.), The innate mind: Structure and
content. New York: Oxford University Press.
The evolutionary genetics of personality
55
Turelli, M., & Barton, N. H. (2004). Polygenic variation maintained by balancing selection:
Pleiotropy, sex-dependent allelic effects and GxE interactions. Genetics, 166, 1053-1079.
Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current
Directions in Psychological Science, 9, 160-164.
Turkheimer, E., & Gottesman, I.I. (1991). Is H2 = 0 a null hypothesis anymore? Behavioral
and Brain Sciences, 14, 410–411.
van der Maas, H. L. J., Dolan, C. V.; Grasman, R. P. P. P., Wicherts, J. M., Huizenga, H. M.,
& Raijmakers, M. E. J. (2006). A dynamical model of general intelligence: The positive
manifold of intelligence by mutualism. Psychological Review, 113, 842-861.
van Oers, K., de Jong, G., van Noordwijk, A. J., Kempenaers, B., & Drent, P. J. (2005).
Contribution of genetics to the study of animal personalities: A review of case studies.
Behaviour, 142, 1185-1206.
Vandenberg, S. G. (1972). Assortative mating, or who marries whom? Behavior Genetics, 2,
127-157.
Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D., & Van Tienderen,
P. H. (1995). Adaptive phenotypic plasticity: Consensus and controversy. Trends in
Ecology and Evolution, 10, 212-217.
Wang, E., Ding, Y. C., Flodman, P., Kidd, J. R., Kidd, K. K., Grady, D. L., et al. (2004). The
genetic architecture of selection at the human dopamine receptor D4 (DRD4) gene locus.
American Journal of Human Genetics, 74, 931-944.
West-Eberhard, M. J. (2003). Developmental plasticity and evolution. Oxford: University
Press.
Yamagata, S., Suzuki, A., Ando, J., Ono, Y., Kijima, N., Yoshimura, K., et al. (2006). Is the
genetic structure of human personality universal? A cross-cultural twin study from North
America, Europe, and Asia. Journal of Personality and Social Psychology, 90, 987-998.
Zhang, X.-S., & Hill, W. G. (2005). Genetic variability under mutation selection balance.
Trends in Ecology and Evolution, 20, 468-470.
The evolutionary genetics of personality
56
Footnotes
1
: Another domain of heritable personality differences for which strong assortative mating
exists are some social attitudes, like conservatism or religiosity (Lykken & Tellegen, 1993;
Eaves et al., 1999). However, unlike the basic personality traits and abilities we treat in this
article, these attitudes must be regarded as complex developmental outcomes of GxE
interactions (Eaves et al., 1999, pp. 77-78). Another noteworthy difference between attitudes
and fitness-related traits like intelligence and attractiveness is that there seems to be no
universal consensus in either sex on the desired attitudes of an ideal mate. It is thus
implausible that competitive mating market dynamics cause assortative mating for attitudes
in a similar way as they do for fitness components. Instead, social homogamy (i.e., mate
search within the own peer group that tends to share similar attitudes) and later dyadic
assimilation appear to be more promising explanations.
The evolutionary genetics of personality
Acknowledgements
We would like to thank three anonymous reviewers for their helpful comments.
57
Table 1
A comparison of evolutionary genetic mechanisms for the maintenance of genetic variation and empirical predictions for affected traits
Genetic variation is due to…
Selective neutrality
Mutation-selection-balance
Balancing selection
…mutations that are not affected by
…an accumulation of many old and
…polymorphisms that are maintained by
selection because their
new, mildly harmful mutations that
selection because the fitness pay-off
phenotypic effect is unrelated to
selection has not yet wiped out of
of their phenotypic effects varies
fitness in any environment.
the population.
across environments.
Predictions for an affected trait:
Number of genetic loci (mutational target size)
Number of polymorphic loci (QTLs)
Average gene effect on trait
Prevalence of polymorphisms
Relation to fitness
Average fitness across environments
Additive genetic variance (VA)
Ratio non-additive to total genetic variance (Dα)
Environmental variance (VE)
Expression dependent on overall condition
Inbreeding depression / heterosis effects
Average social evaluation / sexual attractiveness
(no prediction)
very large
medium
likely small
large
small
(no prediction)
small
medium
intermediate
rare
mostly intermediate
neutral
unidirectional
contingent on environment
equal
unequal
approximately equal
(no prediction)
large
medium
small
medium
high
(no prediction)
large
medium
no
yes
no
weak or none
strong
weak
neutral
strong unidirectional favouritism
weaker, conditional favouritism
Figure captions
Figure 1: The watershed model of genetic variation
Mutations at specific loci (1a, 1b) disrupt narrowly defined mechanisms such as
dopaminergic regulation in the prefrontal cortex (2b). This and other narrowly defined
mechanisms contribute noise to more broadly defined mechanisms, such as working
memory (3c). Working memory in conjunction with several other mechanisms (3a, 3b, 3d)
affect phenotypically observable phenotypes, such as cognitive ability (4). If enough noise is
present in the upstream processes, specific behavioural distortions may arise, such as mild
mental retardation. All tributaries eventually flow into fitness. (Reproduced from Cannon &
Keller, 2005, with permission from www.annualreviews.org.)
Figure 2: Two examples for individual reaction norms
Both figures show the individual reaction norms of three genotypes (A-C) along a continuous
environmental dimension. The trait in Figure 2a has simple reaction norms, where all
genotypes react linearly to environmental changes and differ only in their slope. The trait in
Figure 2b has complex reaction norms, where genotype C reacts linearly and genotypes A
and B react non-linearly in different ways. This leads to different rank orders of reaction
strength at point X, Y, and Z on the environmental dimension, implying the absence of
structural pleiotropy. (Figure 2b is redrawn after van Oers et al., 2005.)
Figure 3: An integrative model of the evolutionary genetics of personality
Note: Mut.: mutation, CD: condition-dependency, RN: reaction norm
The evolutionary genetics of personality
61
Figure 1
1a
2a
2b
3b
3c
3a
3d
4
1b
The evolutionary genetics of personality
Figure 2
62
Figure 3